TW474090B - Speech-enabled information processing - Google Patents
Speech-enabled information processing Download PDFInfo
- Publication number
- TW474090B TW474090B TW89114557A TW89114557A TW474090B TW 474090 B TW474090 B TW 474090B TW 89114557 A TW89114557 A TW 89114557A TW 89114557 A TW89114557 A TW 89114557A TW 474090 B TW474090 B TW 474090B
- Authority
- TW
- Taiwan
- Prior art keywords
- voice
- information
- user
- patent application
- page
- Prior art date
Links
Landscapes
- Telephonic Communication Services (AREA)
Abstract
Description
474090 A7 _ B7 五、發明說明(1 ) 交互參考相關申請案 (請先閱讀背面之注意事填寫本頁) 本申請案先於1 999年6月20日提出之60/144,609美國專利 臨時申請案,名稱爲"SPEECH-ENABLED CALL AND INFORMATION PROCESSING SYSTEM”。 技術領域 本發明與電信有關,更明確地說,與互動式語音應用有 、 關。 發明背景 運用電腦的語音處理系統在各方面的應用愈來愈普遍。 ί i. 某些語音處理系統提供系統與呼叫者/使用者間互動式語音 回應(I V R ) 。I V R系統的應用例包括自動化隨身攜帶 之個人記事簿及客服應用。客服應用包括協助呼叫者得到航 空公司的飛航資訊或訂位,或股票喊價系統。 . 經濟部智慧財產局員工消費合作社印製 某些顧客服務也經由稱爲網際網路的電腦式全球包封-交換網路,特別是經由全球資訊網(nthe web")使用全球資 訊網頁(” w e b p a g e s")組成網站。這些網站典型上包括一” 首頁”,內含某些資訊,並可鏈結到網站中能提供更多資訊 及/或服務的其它網頁。各公司的網頁允許使用者取得公司 的資訊j存取各人的目錄,以及得到其它資訊’例如股票喊 價或飛航資訊,或服務,例如購物(例如C D唱片)或服務 (例如機票)。很多網站包含相同類型的網頁供使用者選擇 ,例如公司資訊、公司的型錄、有關公司的最新消息、以及_ 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) 474090 A7 B7 五、發明說明(2 ) 可提供給使用者的產品/服務。網頁可以使用網路瀏覽器導 覽,典型的導覽工具有”上一頁”、”下一頁”及”首頁” 發明槪述 一般言之,本發明的一態樣是提供一種交談式語音系統 V / ,包括一埠,被架構成接收來自使用者的呼叫,並在系統與 使用者間提供一通信鏈結,記憶體,其內儲存有個人的目錄 ί. ......... - .............. · 資Μ广包括複數個人的指示以及與每一個人有關的線路資訊 ,用以將呼叫安排到複數個人中所選擇的人,記憶體中也儲 存有與互動式語音系統相關之公司的公司資訊,以及語音單 元,與埠及記憶體耦合,並被架構成傳送第一聲音資訊給璋 ,用以提示使用者對系統說話,語音單元也被架構成可經由 埠接收使用者的語音,以識別使用者的語音,並根據所辨識 的使用者語音執行動作,語音單元進一步被架構成能按照儲 存在記憶體中的公司資訊將第二聲音資訊傳送給埠。 本發明的實施包括以下一或多種特徵。 * 語音單元被架構成是以至少部分與網路類似的格式傳送 語音。語音單元被架構成,反應被語音單元識別之使用者的 請求,根據請求,提供儲存在記憶體中的資訊,並根據與該 人有關的線路資訊將呼叫安排到使用者之請求所指示的人。 儲存在記憶體中的部分公司資訊,按照複數個資訊種類在各 資訊頁中相互相關,包括如何與公司聯絡。語音單元被架構 成如果使用者的語音是在最近爲語音單元所存取之資訊頁的. (請先閱讀背面之注意事$填寫本頁) i裝 丨線」 經濟部智慧財產局員工消費合作社印製 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) -5 - 474090 A7 B7 經濟部智慧財產局員工消費合作4J. 五、發明說明(3 ) 用語範圍內,則遵照使用者的語音。資訊的種類進一步包括 v ' 有關公司位置、公司產品(如果有)以及公司所提供的服務 (如果有)。儲存在記憶體中的公司資訊包括可從公司網站 取得的資訊。記憶體及語音單元被架構成傳送給使用者之公 司資訊的組織與提供在公司網站上之公司資訊的組織不同。 語音單元被架構成反應來自使用者的語音命令存取資訊頁, 這些語音命令的功能與網路瀏覽器所提供的相關。這些命令 包括”上一頁”、”下一頁”及”首頁”。 語音單元被架構成經由使用者語音的指示執行交易。 .....-: 系統進一步包括語音應用監視器,被架構成監視語音單 元的動作與來自使用者之對應的進入語音。語音單元被架構 成將對談資料儲存到記憶體中,以指示下列至少其中之一: 使用者的語音;使用者的語音是否經識別後被接受,語音單 元是否採取動作,是採取何種動作;以及,使用者之語音的 信賴度是否低於一既定門檻;語音應用監視器被架構成瀚告 u.' 、 - 語音單元所儲存之對談資料的指示。語音應用監視器經由網 際網路耦合到記憶體。 * 語音單元被架構成執行至少一種使用者語音的岐義消除 ,並確認使用者的語音。、 系統進一步包括一控制單元與記憶體耦合,並被架構成 接收系統外部的控制信號,並反應控制信號修改記憶體中的 資訊。控制單元被架構成可將資訊加入記憶體」可刪除記憶 體中的資訊,以及更改記憶體的資訊。、 語音單元進一步被架構成傳送資訊給使用者以提示使用_ (請先閱讀背面之注意事填寫本頁)474090 A7 _ B7 V. Description of the invention (1) Cross-reference to related applications (please read the notes on the back first and fill out this page) This application was filed before the US Patent Provisional Application 60 / 144,609 filed on June 20, 999 The name is "SPEECH-ENABLED CALL AND INFORMATION PROCESSING SYSTEM". TECHNICAL FIELD The present invention relates to telecommunications, and more specifically, it relates to interactive voice applications. BACKGROUND OF THE INVENTION Application of computer-based voice processing systems in various aspects More and more common. Ί i. Some voice processing systems provide interactive voice response (IVR) between the system and callers / users. Application examples of IVR systems include automated personal notebooks and customer service applications. Customer service applications include Assist callers to get airline flight information or reservations, or stock outcry systems.. Intellectual Property Bureau, Ministry of Economic Affairs, Employee Consumer Cooperative. Certain customer services are also printed via computerized global encapsulation-exchange called the Internet. The internet, in particular via the global web (nthe web ") using global web pages ("webpages ") station. These sites typically include a "homepage" with some information and links to other pages on the site that provide more information and / or services. Each company's webpage allows users to obtain company information, j to access their directories, and other information 'such as stock bargains or flight information, or services such as shopping (such as CD records) or services (such as airline tickets). Many websites contain the same type of web pages for users to choose from, such as company information, company catalogs, latest news about the company, and _ This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) 474090 A7 B7 V. Description of Invention (2) Products / services that can be provided to users. The web page can be navigated by using a web browser. Typical navigation tools are "previous", "next" and "home" Invention Description Generally speaking, one aspect of the present invention is to provide a conversational voice The system V / includes a port, and the frame is configured to receive calls from users, and provides a communication link between the system and the users, a memory, which stores a personal directory ί ............ ..-.............. The information includes a plurality of personal instructions and line information related to each person, which is used to arrange the call to the selected person among the plurality of individuals, and remember It also stores the company information of the company related to the interactive voice system, as well as the voice unit, which is coupled to the port and the memory, and is framed to send the first sound information to the user to prompt the user to speak to the system. The unit is also structured to receive the user's voice through the port to recognize the user's voice and perform actions based on the recognized user's voice. The voice unit is further structured to enable the first Second voice Information is sent to the port. Implementations of the invention include one or more of the following features. * The voice unit is structured to transmit voice in a format at least partially similar to the Internet. The voice unit is constructed to respond to the request of the user identified by the voice unit, provide information stored in the memory according to the request, and arrange the call to the person indicated by the user's request according to the line information related to the person . Part of the company information stored in memory is related to each other on various information pages according to multiple types of information, including how to contact the company. The voice unit is structured if the user's voice is on the information page recently accessed by the voice unit. (Please read the note on the back first to complete this page) i Install 丨 Online ”Consumer Cooperatives, Intellectual Property Bureau, Ministry of Economic Affairs The printed paper size is in accordance with China National Standard (CNS) A4 (210 X 297 mm) -5-474090 A7 B7 Consumer cooperation between employees of the Intellectual Property Office of the Ministry of Economic Affairs 4J. V. Description of invention (3) Within the scope of the term, follow User voice. The types of information further include v 'about company location, company products (if any), and services provided by the company (if any). Company information stored in memory includes information available from the company's website. Memory and voice units are organized differently from organizations that provide company information to users than organizations that provide company information on company websites. The voice unit is structured to access the information page in response to voice commands from the user, and the functions of these voice commands are related to those provided by the web browser. These commands include "Previous", "Next", and "First Page". The voice unit is structured to execute a transaction via a user's voice instruction. .....-: The system further includes a voice application monitor, which is constructed to monitor the actions of the voice unit and the corresponding incoming voice from the user. The voice unit is structured to store the conversation data in the memory to indicate at least one of the following: the user's voice; whether the user's voice is recognized after being recognized, whether the voice unit takes action, and what action is taken ; And, whether the user's voice reliability is lower than a predetermined threshold; the voice application monitor is framed to constitute an indication of the conversation information stored in the voice unit u. ',-Voice unit. The voice application monitor is coupled to memory via the Internet. * The voice unit is constructed to perform at least one disambiguation of the user's voice and confirm the user's voice. The system further includes a control unit coupled to the memory, and is constructed to receive control signals external to the system, and respond to the control signals to modify information in the memory. The control unit is structured to add information to the memory "to delete information in the memory and change information in the memory. 、 The voice unit is further framed to send information to the user to prompt the use of _ (please read the note on the back first and fill in this page)
IIU -丨裝 i訂· •線_ 度適用中國國家標準(CNS)A4規格(210 X 297公釐) -6- 474090 Α7 Β7 經濟部智慧財產局員工消費合作社印製 五、發明說明(4) 者提供消除關於個人岐義的資訊,並使用消除岐義的資訊消 除使用者想要聯絡之人在複數個人中的岐義。 V . 廣義言之,在另一態樣中,本發明提供一種電腦程式產 品,包括電腦-可讀取的指令,用以致使電腦反應接收自使 , i 用者的呼叫,與使用者建立一通信鏈結,以從儲存有個人目 錄資訊的記憶體中擷取資訊,包括複數個人的指示及與每一 個人相關的線路資訊/用以將呼叫安排到複數個人中所選擇 人,記憶體中也儲存有與互動式語音系統相關之公司的公司 資訊,傳送第一聲音資訊給使用者以提示使用者說話,接收 使用者的語音,、識別來自使用者的語音,根據所識別之使用 者的語音執行動作,並按照儲存在記憶體中的公司資訊,傳 送第二聲音資訊給使用者。 本發明的實施包括以下一或多種特徵。~ 致使電腦傳送第二聲音資訊的指令致使電腦以至少部分 與網路類似的格式傳送第二聲音資訊。用以致使電腦傳、送第 二聲音資訊的指令致使電腦反應已由電腦識別之使用者的請 求,按照請求,提供儲存在記憶體中的資訊,電腦程式Μ品 進一步包括用以按照與個人相關的線路資訊致使電腦將呼叫 安排到請求所指示之個人的指令。資訊按照複數個既定的資 訊種類按頁儲存在記憶體中,以及,用以致使電腦識別使用 者語音的指令,致使電腦使用與目前之語音頁相關的字彙識 別使用者的語音。儲存在記憶體中的公司資訊包括可在公司 網站上取得的資訊,以及,致使電腦將第二聲音資訊傳送給 使用者的指令致使電腦傳送給使用者之第二聲音資訊的組織_ 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) (請先閱讀背面之注意事'8|||?填寫本頁) 裝 '1訂· -丨線· 474030 A7 B7 經濟部智慧財產局員工消費合作社印製 五、發明說明(5 ) 與提供在公司網站上之公司資訊的組織不同。致使電腦擷取 k"" 資訊的指令致使電腦反應來自使用者的語音命令擷取資訊, 這些語音命令的功能與網路瀏覽器所提供的相關。這些命令 包括”上一頁”、”下一頁”及”首頁” 〇 「 電腦程式產品進一步包括致使電腦執行使用者語音所指 示之交易的指令。、 電腦程式產品進一步包括致使電腦將對談資料儲存到記 憶體中,用以指示下列至少其中之一:使用者的語音;使用 者的語音是否經識別後被接受,電腦是否採取動作,是採取 何種動作;以及,使用者之語音的信賴度是否低於一既定門 檻,以及報告所儲存之對談資料的指示。 電腦程式產品進一步包括致使電腦根據識別使用者語音 • * 的嘗試執行動作的指令。 電腦程式產品進一步包括致使電腦接收一控制信號並反 應控制信號修改記憶體中資訊內含的指令。致使電腦修改記 憶體中資訊內容的指令包括將資訊加入到記憶體中的指令, 從記憶體中刪除資訊的指令,以及改變記憶體之資訊的指令 二 0 電腦程式產品進一步的指令包括致使電腦:傳送資訊給 使用者以提示使用者提供關於消除個人岐義的資訊,並使用 V I. 消除岐義的資訊消除使用者想要聯絡之人在複數個人中的岐 義。 廣義言之’在另一態樣中,本發明提供一種經由互動式 語音應用與使用者結合的方法,該方法包括接收來自使用者 、一- - , (請先閱讀背面之注意事?^填寫本頁) 裝 卜訂: -丨線‘IIU-丨 booking and ordering • The line _ degree is applicable to the Chinese National Standard (CNS) A4 specification (210 X 297 mm) -6- 474090 Α7 Β7 Printed by the Intellectual Property Bureau of the Ministry of Economic Affairs, Consumer Consumption Cooperative V. Invention Description (4) The author provides disambiguation information about the individual, and uses the disambiguation information to disambiguate among multiple individuals who the user wants to contact. V. Broadly speaking, in another aspect, the present invention provides a computer program product including a computer-readable instruction for causing the computer to respond to receiving a call from the user, i. Communication link to retrieve information from memory storing personal directory information, including instructions from multiple individuals and line information related to each individual / to arrange calls to selected individuals among multiple individuals, also in memory Stores company information of the company related to the interactive voice system, sends the first sound information to the user to prompt the user to speak, receives the user's voice, recognizes the voice from the user, and according to the recognized user's voice Perform the action and send the second sound information to the user according to the company information stored in the memory. Implementations of the invention include one or more of the following features. ~ The instructions that cause the computer to send the second sound information cause the computer to send the second sound information in a format at least partially similar to the network. The instructions for causing the computer to send and send the second sound information cause the computer to respond to the request of the user identified by the computer, and provide the information stored in the memory in accordance with the request. The computer program M product further includes instructions for personally related The line information caused the computer to schedule the call to the individual instructed by the request. The information is stored in the memory in pages according to a plurality of predetermined information types, and the instructions for causing the computer to recognize the user's voice cause the computer to use the vocabulary associated with the current voice page to recognize the user's voice. Company information stored in memory includes information available on the company's website and an organization that causes the computer to send the second sound information to the user and causes the computer to send the second sound information to the user_ This paper standard Applicable to China National Standard (CNS) A4 specification (210 X 297 mm) (Please read the note on the back first '8 |||? Fill in this page) Pack' 1 order ·-丨 · 474030 A7 B7 Intellectual Property of the Ministry of Economic Affairs Printed by the Bureau ’s Consumer Cooperatives 5. The Invention Description (5) is different from the organization that provides company information on the company website. The instructions that cause the computer to capture k " " information cause the computer to respond to the user's voice commands to retrieve information. The functions of these voice commands are related to those provided by the web browser. These commands include "previous", "next" and "home" 〇 "Computer program products further include instructions that cause the computer to perform transactions indicated by the user's voice." Computer program products further include information that causes the computer to talk Stored in memory to indicate at least one of the following: the user's voice; whether the user's voice is recognized after being recognized, whether the computer takes action, and what action it takes; and, the user's voice trust Whether the degree is below an established threshold, and instructions for reporting the stored conversation data. The computer program product further includes instructions that cause the computer to perform an action based on an attempt to recognize the user's voice • *. The computer program product further includes the computer that receives a control Signals and control signals to modify the instructions contained in the information in the memory. The instructions that cause the computer to modify the content of the information in the memory include instructions to add information to the memory, instructions to delete information from the memory, and changes to the memory. Information on instruction 2 0 computer program product further The instructions include causing the computer to: send information to the user to prompt the user to provide information about disambiguation of the individual, and use VI I. Disambiguation Information to disambiguate among multiple individuals who the user wants to contact. Saying 'In another aspect, the present invention provides a method for combining with a user via an interactive voice application. The method includes receiving from a user, a--, (Please read the note on the back first? ^ Fill this Page) Bookbinding:-丨 '
_ _· -_ϋ 1 IWTI 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) -8 - 474090 A7 B7 五、發明說明(6 ) 的進入呼叫,建立與使用者的通信鏈結,擷取語音所指示的 ,' > ; Ll (請先閱讀背面之注意事t填寫本頁) 部分儲存資料用以呈現給使用者,以及,以與網路類似的形 式將部分的儲存資料以語音呈現給使用者。 本發明的實施包括以下一或多種特徵。 所儲存的資訊是按照在每一個對應群組中指示資料內含 的相關標題儲存在群組中,以及,呈現包括以語音將所儲存 之部分資料的標題傳送給使用者。該方法進一步包括接收使 用者的語音,將使用者的語音轉換成使用者之語音的電氣指 示,按照電氣指示擷取儲存資料的另一部分,將儲存資料的 另一部分呈現給使用者包括以語音將另一部分儲存資料的標 題傳送給使用者。使用者的語音是另一部分儲存資料的標題 。使用者之語音的指示是另一部分儲存資料之標題的指示。 語音的指示是儲存資料之另一部分之標題之同義的指示。使 用者的語音包括與網路類似的導覽命令。與網路類似的導覽 命令選用自:”上一頁”、”下一頁”、”首頁”、”到,, 及”說明” 。 經濟部智慧財產局員工消費合作社印製 所儲存的資料按照資料的內容群組,以及,呈現包插傳 送一語音指不給部分儲存資料之資料內容的使用者,指示中 包括”頁”這個字。 廣義言之,在另一態樣中,本發明提供一監視系統,用 以監視至少一個語音應用系統,:監視系統包括一電腦網路通 信線路,以及一監視單元與語音應用系統耦合並耦合到電 腦網路通信線路,並被架構成從至少一個語音應用系統經由 電腦網路通信線路接收資料,處理與語音應用系統相關之呼 -9 - 本紙張尺度適用t國國家標準(CNS)A4規格(210 X 297公釐) 474030 A7 Β7 五、發明說明(7 ) 叫有關之指示的呼叫記錄,並產生與呼叫有關之指示的報告 指示。、 本發明的實施包括以下一或多種特徵。 監視單元經由電腦網路通信線路耦合到語音應用系統, 且監視單元的位置與至少一個語音應用系統間有一段距離。 電腦網路連接經由網際網路耦合到至少一個語音應用系統。 監視單元被架構成存取儲存在至少一個語音應用系統中之呼 叫記錄的記錄。監視單元經由電腦網路通信線路及網際網路 稱合到複數個分散的語音應用系統,且裤架構成經由網路通 信線路接收每一個語音應用系統的資料,處理與每一個語音 應用系統相關之呼叫事件的記錄,並爲每一個語音應用系統 產生與呼叫相關之指不的指不報告。 監視單元被架構成傳送信號給至少一個語音應用系統, 以改變至少一個語音應用系_統的操作。信號適合致使至少一 個語音應用系統從故障中的通信線路有效地恢復忙碌。、信號 適合致使至少一個語音應用系統的服務重新開始。信號適合 致使補綴的組態檔插入至少一個語音應用系統中的組態續。 監視單元被架構成能產生所選呼叫事件之頻率的指示。 監視單元被架構成能產生關於所選呼叫事件的警告。該 警告ί日不所运呼叫事件的特徵偏離超過該特徵之既定參考値 一既定量。 監視單元與語音應用系統的配置相互靠近。 本發明的各態樣提供以下一或多項優點。人們可以使用 電話或其它類似的裝置存取某公司的資訊並從其得到服務。 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) H _ϋ ϋ n n 1_ϋ n l I I 0 ί ϋ (請先閱讀背面之注意事III填寫本頁) 訂: 經濟部智慧財產局員工消費合作社印製 10 474090 A7 B7 五、發明說明(8 ) (請先閱讀背面之注意事填寫本頁) 資訊及/或服務可利用聲音的型式提供及存取、,且格式與網 站相同、且可不經由電腦存取。呼叫者可以經由一般說話的 語言存取資訊及服務。以聲音互動的語音格式提供相同的資 訊及/或服務,對公司在網站或其它資訊投資的增殖上具有 一槓桿作用。呼叫者可以使用一般網路瀏覽器所使用的命令 導覽公司的資訊及/或服務。互動語音的性能可以被監視。 監視可以在遠端執行,例如經由網際網路。多個互動式語音 回應系統可以在遠方監視。一或多個互動式語音回應系統可 以在遠方控制。遠方控制可以包括建立及/或修改資料,例 如用來識別語音及/或回應語音或其它聲音所執行之動作的 組態參數及資料。 從以下的圖式、詳細描述及申請專利範圍,將可完全明 瞭本發明的這些及其它優點連同發明本身。 圖式簡單說明 ^ 圖1是按照本發明之語音系統的簡單方塊圖。 經濟部智慧財產局員工消費合作社印製 圖2是經由網路鏈結連接到伺服器之電腦系統的簡車方 塊圖。 圖3是圖1中所示I V R系統的簡單方塊圖,以及一類 比線路、一單郵件傳送協定伺服器以及一電傳伺服器。 圖4是連接到多個I V R系統之分析及/或報告服務的 簡單方塊圖。 圖5是按照本發明之交談語音法的簡單方塊圖。 圖6是圖5中所示安排線路方法的流程圖。 -11- 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) 474090 A7 B7 經濟部智慧財產局員工消費合作社印製 五、發明說明(9) 圖7是圖5中所示資訊擷取方法的流程圖。 圖8是圖5中所示交易處理方法的流程圖。 圖9是報告及分析交談對話之方法的流程圖。 圖10是圖3所示之引擎系統的簡單方塊圖。 元件對照表 -1 丄 〇 互動式語音系統 1 2 使用者及/或呼叫者 1 4 公用交換式電話網路_ 1 6 互動式語音回應系統 1 8 簡單郵件傳送協定伺服器 2 〇 防火牆 2 2 網際網路 2 4 分析/報告服務 2 6 電話 2 8 傳真機 3 〇 引擎系統 3 2 管理系統 3 4 組態與記錄系統 5 〇 電腦系統 5 2 匯流排 5 4 處理器 5 6 主記憶體 5 8 唯讀記憶體 -------*-------裝--- (請先閱讀背面之注意事HI填寫本頁) · 線」 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) -12- 474090 A7 B7 經濟部智慧財產局員工消費合作社印製 五、發明說明(ίο) 〇 4 7〇 3 6 3 8 4〇 4 2 4 4 8 0 3 0 0 3 0 2 3 0 4 3 0 6 3 0 8 3 1〇 8 6 8 8 9〇 9 2 9 4 9 6 儲存裝置 顯示器 輸入裝置 游標控制 通信介面 網路鍵結 遙控系統 支援系統 監視介面系統 網路應用伺服器 應用伺服器邏輯 執行引擎 語音處理單元 語音識別器 提示單元 記錄單元 服務邏輯執行環境 硬體 記錄儲存區 資料庫儲存區 一般儲存區 文字-到-語音特徵 記錄轉換器 傳真特徵 ----------------裳--- (請先閱讀背面之注意事m填寫本頁) 訂· -·線'_ _ · -_Ϋ 1 IWTI This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) -8-474090 A7 B7 V. Incoming call of invention description (6), establish communication link with user Finally, capture the voice instructions, '>; Ll (please read the note on the back first to fill out this page) Part of the stored data is presented to the user, and part of the stored is similar to the Internet The data is presented to the user in voice. Implementations of the invention include one or more of the following features. The stored information is stored in the group according to the relevant title contained in the instruction data in each corresponding group, and the presentation includes transmitting the title of a part of the stored data to the user by voice. The method further includes receiving a user's voice, converting the user's voice into an electrical instruction of the user's voice, retrieving another portion of the stored data in accordance with the electrical instruction, and presenting the other portion of the stored data to the user, including the The title of the other part of the stored data is sent to the user. The user's voice is the title of another part of the stored data. The instructions of the user's voice are instructions for another part of the title of the stored data. Voice instructions are synonymous instructions for the title of another part of the stored data. The user's voice includes navigation commands similar to the web. Navigation commands similar to the Internet are selected from: "Previous page", "Next page", "Home page", "To", and "Description". The printed data stored by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs According to the content group of the data and the presentation packet transmission, a voice refers to a user who does not partially store the data content of the data, and the instruction includes the word "page". Broadly speaking, in another aspect, the present invention A monitoring system is provided for monitoring at least one voice application system. The monitoring system includes a computer network communication line, and a monitoring unit is coupled to the voice application system and coupled to the computer network communication line, and is constructed from at least one The voice application system receives data through the computer network communication line and processes calls related to the voice application system. -9-This paper size is applicable to the national standard (CNS) A4 specification (210 X 297 mm) 474030 A7 B7. 5. Description of the invention (7) Call the related call records, and generate report instructions for call related instructions. The implementation of the present invention includes one or more of the following Features. The monitoring unit is coupled to the voice application system via a computer network communication line, and the location of the monitoring unit is at a distance from at least one voice application system. The computer network connection is coupled to the at least one voice application system via the Internet. The monitoring unit The frame constitutes a record that accesses the call records stored in at least one voice application system. The monitoring unit is connected to a plurality of decentralized voice application systems via a computer network communication line and the Internet, and the pants rack constitutes communication via the network The line receives the data of each voice application system, processes the record of call events related to each voice application system, and generates a call-related finger or finger report for each voice application system. The monitoring unit is framed to form a transmission signal Give at least one voice application system to change the operation of the at least one voice application system. The signal is suitable to cause the at least one voice application system to effectively resume busy from the faulty communication line. The signal is suitable to cause the service of the at least one voice application system to restart Start. Signal fit Insert the patched configuration file into the configuration of at least one voice application system. The monitoring unit is constructed to generate an indication of the frequency of the selected call event. The monitoring unit is constructed to generate a warning about the selected call event. The WARNING: The characteristics of unforeseen call events deviate more than the established reference value of the characteristic. The configuration of the monitoring unit and the voice application system are close to each other. Aspects of the present invention provide one or more of the following advantages. People can use the telephone Or other similar devices to access the information of a company and get services from it. This paper size applies Chinese National Standard (CNS) A4 (210 X 297 mm) H _ϋ ϋ nn 1_ϋ nl II 0 ί ϋ (Please read first Note on the back III fill in this page) Order: Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 10 474090 A7 B7 V. Description of the invention (8) (Please read the note on the back first and fill in this page) Information and / or services are available It is provided and accessed using a sound type, in the same format as the website, and can be accessed without a computer. Callers can access information and services in a common language. Providing the same information and / or services in a voice-interactive voice format can have a leverage on the company's proliferation of websites or other information investments. Callers can use commands used by normal web browsers to navigate company information and / or services. The performance of interactive speech can be monitored. Surveillance can be performed remotely, for example via the Internet. Multiple interactive voice response systems can be monitored from a distance. One or more interactive voice response systems can be controlled from a distance. Remote control may include creating and / or modifying data, such as configuration parameters and data used to identify speech and / or respond to actions performed by speech or other sounds. These and other advantages of the present invention, as well as the invention itself, will be fully apparent from the following drawings, detailed description, and patent application scope. Brief description of the drawings ^ Figure 1 is a simple block diagram of a speech system according to the present invention. Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs Figure 2 is a block diagram of a computer system connected to a server via a network link. Fig. 3 is a simple block diagram of the I V R system shown in Fig. 1, and an analog line, a single mail transfer protocol server, and a telex server. Figure 4 is a simple block diagram of an analysis and / or reporting service connected to multiple IVR systems. FIG. 5 is a simple block diagram of a conversational speech method according to the present invention. FIG. 6 is a flowchart of a method for arranging lines shown in FIG. 5. FIG. -11- This paper size is in accordance with Chinese National Standard (CNS) A4 (210 X 297 mm) 474090 A7 B7 Printed by the Consumers ’Cooperative of Intellectual Property Bureau of the Ministry of Economic Affairs 5. Description of the invention (9) Figure 7 is shown in Figure 5 Flow chart of information retrieval method. FIG. 8 is a flowchart of the transaction processing method shown in FIG. 5. FIG. 9 is a flowchart of a method of reporting and analyzing a conversation. FIG. 10 is a simple block diagram of the engine system shown in FIG. 3. Component comparison table-1 丄 〇 Interactive Voice System 1 2 Users and / or Callers 1 4 Public Switched Telephone Network_ 1 6 Interactive Voice Response System 1 8 Simple Mail Transfer Protocol Server 2 〇 Firewall 2 2 Internet Network 2 4 Analysis / Reporting Service 2 6 Telephone 2 8 Facsimile 3 〇 Engine System 3 2 Management System 3 4 Configuration and Recording System 5 〇 Computer System 5 2 Bus 5 4 Processor 5 6 Main Memory 5 8 Only Read memory ------- * ------- load --- (please read the note on the back HI to fill out this page first) · Line "This paper size applies to China National Standard (CNS) A4 specifications (210 X 297 mm) -12- 474090 A7 B7 Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs V. Invention Description (ίο) 〇 4 7〇3 6 3 8 4〇4 2 4 4 8 0 3 0 0 3 0 2 3 0 4 3 0 6 3 0 8 3 1〇8 6 8 8 9〇9 2 9 4 9 6 Storage device display input device cursor control communication interface network key remote control system support system monitoring interface system network application servo Application server logic execution engine speech processing unit speech recognizer reminder Meta Recording Unit Service Logic Execution Environment Hardware Record Storage Area Database Storage Area General Storage Area Text-to-Voice Feature Record Converter Fax Feature ---------------- Shang --- (Please read the notes on the back first and fill in this page) Order ·-· Line '
If 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公爱) -13- 474030If this paper size is applicable to China National Standard (CNS) A4 (210 X 297 public love) -13- 474030
A7 B7 1〇 12 14 16 2 0 2 2 2 4 2 6 2 8 報告產生器 S吾音調適器 傳真伺服器 類比通信線路 PC Any where 軟體 數據機 系統監視器 提示監視器 調整監視器 互動式語音回應系統 互動式語音回應系統 互動式語音回應系統 防火牆 S Μ T P伺服器 發明詳細說明 (請先閱讀背面之注意事'^填寫本頁) Λ —裝A7 B7 1〇12 14 16 2 0 2 2 2 2 4 2 6 2 8 Report generator SW tone adapter fax server analog communication line PC Any where software modem system monitor prompt monitor adjustment monitor interactive voice response System Interactive Voice Response System Interactive Voice Response System Firewall SM TP Server Invention Detailed Explanation
-丨線I 槪述 經濟部智慧財產局員工消費合作社印製 本發明的實施例提供語音式的資訊處理系統,與現行之 全球資訊網的網站及系統有一定的關係。例如,企業或公司 具有一運用網路的股票交易系統,可以架構一語音式的資訊 處理系統,它允許使用者經由網路使用以電話進入且具有與 企業網站一致之使用者介面的語音式系統,連絡經紀人或詢 問有關交易的狀況〜在本文中所使用的”公司”一詞,包括 可能使用本文所描述之技術的各類實體。實體可以是或不是 14- 本紙張尺度過用中國國家標準(CNS)A4規格(210 X 297公釐) 474030 A7 ___ B7 五、發明說明(12) 一事業,也可以是或不是想要earn profit。因此,”公司” 一詞包括但不限於公司、社團法人、合股公司、私人團體及 個人。使用”公司”,是因爲網站上通常都是使用這個名詞 ’當然,在本文中並非必須使用這個名詞、。 本發明的實施例支援目前可使用網路技術的各種應用, 包括但不限於通信應用、資訊擷取、交易處理。較佳的情況 是’所有這類應用都可經由與網站類似之單一、一致的使用 者介面獲得,包括超連結;使用者可以對數種應用中任何一 種說出一命令,不需顧慮實際上是一或多個伺服:器或系統執 行應用程式。 使用者可以經由呈現在指引式對話格式中的資訊導覽。 在互動對話期間,會有一組對應於使用者所要述說之命令的 選項提供給使用者。例如,使用者可能聽到“您可以說,,與 我們連絡”、”公司資訊”、或”產品””。也可以提供使 用者一小段命令功能的簡介,例如“您可以說”傳真它、”即 可接收到您剛才所聽到之資訊的傳真。”使用導引式對話有 助於限制需要識別的字彙及語音識別的的速度。· ~ 通信應用包括安排呼叫線路,其中,呼叫者說出呼叫所 要到達之關係人的姓名或敘述所要到達的部門。 交易處理應用可包括無收入的支援處理,例如將錢從一 個銀行帳戶轉到另一帳戶。由於典型上企業無法從這類支援 .功能中獲取營利,使用本文所揭示的語音介面及語音式系統 ,代表可以節省大量的處理成本。_ 交易處理應用也包括電子商務或採購交易。因此,本發 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) (請先閱讀背面之注意事填寫本頁) -丨裝-丨 Line I Description Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs The embodiments of the present invention provide a voice information processing system, which has a certain relationship with the current World Wide Web site and system. For example, a business or company has a stock trading system that uses the Internet and can construct a voice-based information processing system that allows users to use a voice-based system that is accessed by telephone over the Internet and has a user interface consistent with the corporate website Contact a broker or inquire about the status of the transaction ~ The term "company" as used in this article includes all types of entities that may use the technology described in this article. The entity may or may not be 14- This paper is scaled to the Chinese National Standard (CNS) A4 specification (210 X 297 mm) 474030 A7 ___ B7 V. Description of the invention (12) A business may or may not want earn profit . Therefore, the term "company" includes, but is not limited to, companies, corporate entities, joint-stock companies, private groups, and individuals. "Company" is used because the term is usually used on websites. Of course, this term is not necessary in this article. The embodiments of the present invention support various applications that currently can use network technology, including but not limited to communication applications, information retrieval, and transaction processing. The better case is that 'all such applications are available through a single, consistent user interface similar to a website, including hyperlinks; users can say a command on any of several applications without having to worry about actually being One or more servers: servers or systems running applications. Users can navigate through information presented in a guided dialogue format. During the interactive dialogue, a set of options corresponding to the commands the user wants to speak is provided to the user. For example, the user may hear "You can say, contact us", "Company Information", or "Product". You can also provide the user with a brief introduction to the command function, such as "You can say" Fax it, " You will receive a fax of the information you just heard. "Using guided conversations helps limit the speed of vocabulary and speech recognition that need to be identified. · Communication applications include arranging call lines where the caller speaks the name of the person to whom the call is to be reached or describes the department to which it is to arrive. . Transaction processing applications can include non-revenue support processing, such as transferring money from one bank account to another. Because businesses typically cannot make money from such support. Features, use the voice interface and voice-based methods disclosed in this article. System, representatives can save a lot of processing costs. _ Transaction processing applications also include e-commerce or procurement transactions. Therefore, the paper size of this publication applies to the Chinese National Standard (CNS) A4 specification (210 X 297 mm) (please read the back first) Notes (fill in this page)-丨
A 經濟部智慧財產局員工消費合作社印製 -15- 474090 A7 B7 五、發明說明(13) 明的實施例可以提供進到通用交易處理系統的語音式通路, 用以經由線上貿易系統或傳統的back-office貿易系統進行商 業交易。 交易處理也包括互動對話以使呼叫者爲事件登記。對話 也包括以姓名、住址、傳真號碼等識別個人。對話也包括以 信用卡或類似物得到支付資訊。 應用也包括註冊的使用者得到進入一或多種服務或資訊 的特殊權益,或使註冊的使用者能使用私人的選單、團體折A Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs-15- 474090 A7 B7 V. Description of the Invention (13) The embodiment described can provide a voice-type path to the general transaction processing system for online trading system or The back-office trading system conducts business transactions. Transaction processing also includes interactive conversations to enable callers to register for events. Dialogue also includes identifying individuals by name, address, fax number, and so on. Dialogue also includes obtaining payment information by credit card or the like. Applications also include special rights for registered users to gain access to one or more services or information, or enable registered users to use private menus and group discounts.
'〆 V 扣或"cookies”。應用也包括經由一或多個電路-切換的載波 ,或經由網際網路電話(”網際網路上傳送語音的協定”) 連接,將一個語音處理系統鏈結到其它的語音處理系統。應 用也包括提供網站到一或多個語音處理系統的指標器,因此 ,使用者可以知道何種服務可被語音起動,並迅速地從網站 移動到語音處理系統所提供_的服務。 本發明的實施例也增進對古早伺服器的存取。本發、明的 實施例做爲進入到back-office資料伺服器的前端或網路通口 ,以同樣的方式,網路伺服器可位於古早資料的前方。‘ ...... 、. 本發明的實施例可被架構成與網路伺服器結合,以提供. 與網路伺服器所進行相同之資訊擷取功能或交易一方便的介 面與呈現層。.因此’企業可槓桿他的網路投資。通常在網站 上才能看到的功能,現在也能經由使用一般語言的語音介面 以電話存取,使用者只須說出想要之功能的名稱,例如:與 我們連絡、就業機會、關於我們等。某企業也可以使用語音 起動該企業特有的功能。例如,快遞服務業可以經由它的網_ 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) -16 - (請先閱讀背面之注意事填寫本頁)'〆V bucks or " cookies.' Applications also include linking a voice processing system via one or more circuit-switched carriers, or via an internet phone ("the protocol for transmitting voice over the Internet") connection. To other voice processing systems. Applications also include indicators that provide websites to one or more voice processing systems, so users can know what services can be activated by voice and quickly move from the website to the voice processing system _ Services. The embodiments of the present invention also enhance access to the ancient server. The embodiments of the present invention and the present invention are used as the front-end or network port to access the back-office data server. In the same way, The web server can be located in front of the ancient and early data. '... The embodiments of the present invention can be combined with a web server to provide the same information as that performed by the web server. A convenient interface and presentation layer for capturing functions or transactions. So 'a company can leverage his online investment. Features that are usually seen on the website can now also be accessed via voice in common languages Face to face access by phone, the user only needs to say the name of the desired function, such as: contact us, employment opportunities, about us, etc. A company can also use voice to activate the company's unique functions. For example, the express service industry You can go through its net _ This paper size is applicable to China National Standard (CNS) A4 specification (210 X 297 mm) -16-(Please read the notes on the back to fill in this page first)
一P 裝 經濟部智慧財產局員工消費合作社印製 474090 A7 B7 五、發明說明(14) 站提供投寄追蹤服務或資費計算。相同的服務可使用本發明 . . 的實施例經由電話存取,無論該項服務是否提供在網站上。 呼叫者只需簡單地回應系統的致候語說出所要的服務名稱。 在這些服務中所提供的資訊,可以即時鏈結到外部的內容提 供者提供資訊,、以及更複雜的動作。 資訊擷取應用可包括非常簡單的資訊更新,例如追蹤已 使用快遞服務交運的包裹,追蹤航空公司的行李,得到銀行 帳戶的餘額等:以及更複雜的動作。 另一項資訊擷取的應用包括提供想要到達企業中某一位 置之呼叫者的行車方向。呼叫者呼叫按照本發明之實施例的 語音式系統,並回應致候語,例如說”公司資訊’方向”等 。回應此,呼叫者被提示以”你來自的方向? ”呼叫者回應 一方向及可識別的點,例如主要道路。回應此’呼叫者聽到 ”我將讀取你的方向”。接著將方向呈現給呼叫者。結果是 ,以語音提供有用的資訊擷取功能。 、 資訊擷取功能也包括擷取新聞稿、資料單、及其它電子 文件,以及經由文字到語音、傳真或其它媒體的電子文#傳 輸。、. 應用可以使用各種方法架構。例如,實施例包括工具、 套裝軟體,以及架構精靈,它可供操作者建立一新的應用, 提供不同的資訊擷取及交易處理功能。 因此,所揭示的實施例可增進電話的應答,使用公司在 全球資訊網上的投資發揮槓桿效應,並提供各種資訊擷取及 交易處理功能的要件或網路通口 ·。本發明的實施例提供互動_ 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) (請先閱讀背面之注意事$填寫本頁) «is 經濟部智慧財產局員工消費合作社印製 -17- 474090 A7 B7 五、發明說明(15) 請先閱讀背面之注意事ϋΐ填寫本頁) 式的語音系統,組織、資訊內容及服務都依循網路式的模式 L使用者可以使用一具電話發出呼叫,並以自然的語言及互 動式語音回應(I V R )系統存取資訊及/或服務。因此, 本發明的實施例允許使用者經由姓名及/或部門將線路安排 到所選擇的公司員工,.並提供以類網站的組織、名詞及命令 存取公司的資訊及交易。、 經濟部智慧財產局員工消費合作社印製 本發明的典型實施例包括一基礎平台及工具組,·以及可 架構、預套裝應用模組的集合.。有了基礎平台及工具組,可 以使用工具組定製一系統,以提供一按個人量身:訂做的互動 式語音應用。:有了可架構、預套裝應用模組的集合,基本上 客戶是購買一套完整的產品,只需做小部分修改,即可將產 品架構成符合客戶所需。如本發明的實施例在I V R系統中 所提供的類網站功能,本發明的實施例稱爲SpeechSiteTM IVR 系統,包括 SpeechSite™ IVR 介面。在 SpeechSite™ IVR 系 統中的語音頁類似網站的網頁,提供資訊及/或服務,、以不 同的語音頁提供不同群組或種類的資訊及/或服務,與網站 通常所提供的資訊及/或服務相同,且是以典型的網站組織 提供。 以下的描述是假設某一公司購買及使用所描述的實施例 。因此,假設所描述的實施例提供與購買之公司相關的資訊 及產品/服務。不過,與公司不同的實體也可以接受。 結構 全系統 -18- 本紙張尺度適用中國國家標準(CNS)A4規格(210 χ 297公爱) 474090 ..f A7 B7 經濟部智慧財產局員工消費合作社印製 五、發明說明(16) 現請參閱_ι1:互·動式語音系統1 0包括使用者及/或 呼叫者12,公用交換式電話網路(PSTN) 14、 I V R系統1 6、簡單郵件傳送協定(S Μ T P )伺服器 1 8、防火牆2 0、網路2 2,在此是網際網路、以及分析 /報告(A / R )服務2 4。如圖所示,系統1 〇中每一組 件間的通信都是雙向的。呼叫者1 2進入電話2 6及傳真機 28。呼叫者12可以經由電話26或傳真機28與. PSTN 14通信,反之亦然。呼叫者12經由PSTN 1 4與I V R系統1 6通信。I V R系統1 6:以導引式對 話格式提示呼叫者1 2以與呼叫者1 2互動,並識別(或至 少嘗試著識別)來自呼叫者1 2的語音。I V R系統1 6也 經由網際網路2 2與A / R服務2 4通信。S Μ T P伺服器 1 8提供I V R系統與網際網路2 2間的介面。防火牆2 0 使用習知的技術保護I V R系統1 6與網際網路2 2間的通 信。I V R系統1 6包括引擎系統3 0、管理系統3 2、及組 態與記錄系統3 4 〇 系統3 0、3 2、3 4處理I V R系統1 6與呼叫奢 1 2間的互動,分別用於架構引擎系統3 0,儲存組態參數 、提示及其它資料,以及記錄與呼叫者1 2間的互動,如以 下更詳細的說明。 百動式語音回應系統 簡介 I V R系統1 6可以使用個人電腦實施彡例如以下的組_ 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) -19- ----;---------—身--- (請先閱讀背面之注意事填寫本頁) 言· Γ 良 474090 Α7 _-_____ Β7 五、發明說明(17 ) 件及/或特徵可做爲實施I V R系統1 6之電腦的一部分: 單處理器的工作站,使用I NTEL®PENT I UM® (請先閱讀背面之注意事ΐϋι填寫本頁) I I I ( N T工作站認證)處理器,時計速率4 5 Ο Μ Η z 或更高;384Mb RAM或更多;9GB的硬碟空間; 以及高速DLT備份系統;連接用的1 O/l 〇 〇 Ethernet 連接及5 6 K數據機;監視器;滑鼠;鍵盤,用以做資料的 顯示、輸入及操作;D41 ESC及D240 SC-T 1電話介面卡;Antares 6000/50數位信號處理器;NT4.0 工作站 service pack 5作業系統;Artisoft® 5.0 Enterprise的環 境;Access或 SQL伺服器;IIS或 Pure Information Services HTTP 伺服器,以及用於使用FTP Services之Windows NT Services的 Microsoft® FTP Services,或 Apache Software Foundation HTTP伺服器;用於文字-到-語音的Lucent單線路執照;PolyPM 或PCAny where程式用於遠端(例如桌面)管理。 經濟部智慧財產局員工消費合作社印製 現請參閱圖2,實施I V R系統1 6的電腦系統5、0包 括匯流排5 2或其它的通信機制,用以傳送資訊,與匯流排 5 2耦合的處理器5 4用以處理資訊。電腦系統5 0也运括 主記億體5 6,如隨機存取記憶體(R A Μ )或其它的動態 儲存裝置,耦合到匯流排5 2用以儲存資訊及供處理器5 4 執行的指令。在處理器5 4執行指令期間,也使用主記憶體 5 6儲存臨時性的變數或其它中間資訊。電腦系統5 0還包 括唯讀記憶體(R〇Μ ) 5 8或其它靜態儲存裝置耦合到匯 流排5 2,用以儲存靜態資訊及處理器5 4使用的指令。儲 存裝置6 0,如磁碟或光碟被架構成儲存資訊及指令,也耦_ -20 - 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公爱) 474090 A7 B7 經濟部智慧財產局員工消費合作社印製 五、發明說明(18) 合到匯流排5 2。、 · 電腦系統5 0經由匯流排5 2耦合到顯示器6 2 ’如陰 極射線管(C R T )/用以顯示資訊給電腦的使用者。輸入 裝置6 4,如包括文數字或其它按鍵的鍵聲/,耦合到匯流排 5 2,用以將資訊及所選擇的命令傳送給處理器5 4。系統 I. 5 〇中其它類型的使用者輸入裝置包括游標控制6 6,如滑 鼠,軌跡球,或游標方向鍵J用以將方向資訊及所選擇的命 令傳遞給處理器5 4,用以控制游標在顯示器6 2上的移動 。此輸入裝置典型上是在兩軸上控制游標,第一:軸例如X, 第二軸例如y,允許裝置在一平面上指示位置。 按照本發明的實施例,電腦系統5 0反應處理器5 4執 行包含在主記憶體5 6中之一或多個指令一或多次,可以產 生語音識別應用。這些指令是從其它的電腦可讀取媒體,如 儲存裝置6 0讀到主記憶體5 6中。處理器5 4順序執行包 含在主記憶體5 6中的指令致使處理器5 4執行本文中.所描 述的方法。在其它實施例中,硬線電路、韌體、任何種類的 硬體組合及/或軟體都可用來實施本發明的實施例。‘ 本文所使用的”電腦可讀取媒體”包括任何可提供指令 給處理器5 4供其執行的媒體。這類媒體的型式很多,包括 但不限於非揮發性媒體、揮發性媒體、.以及傳輸媒體。非揮 發性媒體例如包括光碟或磁碟,如儲存裝置6 0。揮發性媒 體包括動態記憶體^如主記憶體5 6'。傳輸媒體包括同軸電 纜、銅線及光纖纜,包括構成匯流排5 2的線。傳輸媒體也 可以是超音波或電磁(例如光波·)的型式,如資料通信期間 (請先閱讀背面之注意事lit填寫本頁) •i 言 Γ 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) -21 - 474090 Α7 -------_ Β7 五、發明說明(19) 所產生的無線電波及紅外線。、. 電腦可讀取媒體一般的型式例如包括軟式磁碟、硬式磁 碟、磁帶或其它的磁性媒體、CD - ROM、任何其它的光 學媒體、孔卡、紙帶、任何具有孔圖案的其它物理媒體、 RAM、 PROM、 EPROM、 FLASH— EPROM ’任何其它的記憶體晶片或卡匣(例如電氣及/或電磁,包 括光學),如後文中描述的載波,或任何其它的電腦可讀取 媒體。 各種型式的電腦可讀取媒體都可以攜載一或:多個指令的 一或多個序列供處理器5 4執行。例如,起初,指令可以載 在遠端電腦的磁碟上。遠端電腦可以將指令下載到它的動態 記憶體,並使用數據機經由電話線傳送指令。位在電腦系統 5 0的數據機可以從電話線上接收資料,,以及使用紅外線發 射器將資料轉換成紅外線信_號紅外線檢知器可以接收載在 紅外線信號內的資料,並有適當的電路將資料放到匯流、排 5 2上。匯流排5 2可將資料載到主記億體5 6,處理器 5 4可從主記憶體5 6擷取指令並執行。主記憶體5 6命接 收的指令可以在處理器5 4執行前或執行後選擇性地儲存到 儲存裝置6 0中。 電腦系統5 0也包括耦合到匯流排5 2的通信介面6 8 。通信介面6 8提供雙向通信,耦合到網路鏈結7 0再耦合 到S Μ T P伺服器1 8例如.,通信介面6 8可以是整合式 服務數位網路(I S D Ν )卡或數據機.,連接到對應型式的 電話線提供資料通信。在另一例中,通信介面6 8可以是區 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) (請先閱讀背面之注意事$填寫本頁) 言· Γ 經濟部智慧財產局員工消費合作社印製 -22- 474090 A7 經 濟 部 智 慧 財 產 局 員 工 消 費 合 作 社 印 製 五、發明說明(20) 域網路(LAN)卡,提供與相容之LAN間的資 接。v也可以使用無線鏈結。通信介面6 8可以傳送 氣及/或電磁(包括光學)信號,這些信號中載有 類型之資訊的數位資料流。 電腦系統5 0可以經由S Μ T P伺服器1 8 .、 7 0及通信介面6 8送出訊息及接收資料,包括程 如,可以從網際網路2 2 (圖1 )上下載碼,用以 中所描述的語音識別應用。處理器5 4可以在接收 行所接收的碼,及/或儲存在儲存裝置6 0或其它 性記憶體中,供稍後執行。按照此方法,電腦系統 在載波的型式中得到應用碼。 現請參閱圖3,I V R系統1 6包括引擎系統 理系統3 2、組態與記錄系統3 4、遙控系統(R 3 6、支援系統3 8及監視介面系統4 0。如圖3 箭頭所示,這些系統都可雙向通信。此外,遙控系 以與系統3〇、32、34、38及40雙向通信 統3 2負責I V R系統1 6的架構、報告及監視。 3 2包括網路應用伺服器4 2及應用伺服器邏輯4 管理系統3 2要經由網路應用伺服器4 2,例如Η 服器,當然,其它的伺服器也可以接受。應用伺服 由應用伺服器邏輯4 4控制,它是以軟體實施。 料通信連 及接收電 代表各種 網路鏈結 式碼。例 產生本文 到碼時執 的非揮發 5 0可以 3〇、管 S C ) 中的雙向 統3、6可 。管理系 管理某統 4。存取 Τ Τ Ρ伺 益4 2票 管理系統 管理系統3 2負責架構I VR系統1 6的其它組件 管 ----------------裝--- (請先閱讀背面之注意事填寫本頁) 訂· 474030 A7 ___B7 五、發明說明(21) 理系統3 2被架構成存取儲存在組態與記錄系統3 4中的資 訊:並將此資訊當成組態資訊提供給I V R系統1 6的其它 組件。例如管理系統3 2被架構成從組態與記錄系統3 4讀 取組態資料,並將此資訊提供給引擎系統3 0.。傳送給引擎 系統3 0的組態資料中包括用來決定那一個語音頁要被起動 的資料,以及,頁的內容包括提示的位置,要使用何種文法 ,以及要使用那些字彙等資料。 語音頁按照管理系統3 2的組態資訊被群組到語音模組 中。不同的語音模組提供不同種類的資訊及服務:。每一個模 組包含許多語音頁。模組的典型例是語音陪從( SpeechAttendant)、與我們聯絡、以及公司資訊。語音陪從 模組包括公司中人員目錄的頁。與我們聯絡模組包括與公司 連絡的資訊,例如電子郵件地址、郵件地址、街名地址及住 址、傳真號碼及電話號碼。公司資訊模組包括描述公司一般 資訊的頁,例如事業的種類及/或事業所提供的服務、、新聞 、及服務的地理區域。 管理系統3 2也被架構成用以更新/編輯包含在組態與 記錄系統3 4內的資訊。因此,例如,管理系統3 2可以存 取及編輯員工姓名及電話號碼表,公司資訊例如新聞,以及 呼叫流程控制資料。 引擎系統 . 引擎系統3 0負責及被架構成執行I V R系統1 6與呼 叫者1 2 (圖1 )間的語音介面.,以及與電話系統的連接。 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) (請先閱讀背面之注意事'ίΡΡ填寫本頁) lii 經濟部智慧財產局員工消費合作社印製 -24- 474090 A7 B7 五、發明說明(22) (請先閱讀背面之注意事Ιϋ填寫本頁) 因此,、引擎系統3 0連接到P S T N 1 4做雙向通信,並 適合執行與呼叫者1 2的呼叫流程,包括識別語音、播放提 示、擷取資料、安排呼叫線路、消除回應的岐義,確認回應 、選擇要鏈結的頁,以及連接到支援裝置(例如T T S及傳 真)。引擎系統3 0被架構成使用習知的技術識別語音,將 語音剖析成小段,並應用聲音模型。引擎系統3 0的功能使 用麻州波士頓之Speech Works® International Inc.的 D^logModule™語音處理軟體實施。安排呼叫線路、擷取資 訊及交易處理每一種功能是語音引擎3 0嘗試識:別呼叫者的 語音,以及據以採取的動作。 引擎系統3 0也包括執行引擎8 0,被架構成控制引擎 系統3 0的處理。此處理包括擷取及播放提示、語音識別、 以及監視及報告引擎系統3 0與呼叫者1 2間的互動。引擎 系統3 0被儲存在引擎系統3 0或組態與記錄系統3 4或呼 叫者1 2所提供的指令及/或資料控制。 、 經濟部智慧財產局員工消費合作社印製 現請參閱圖1 0,引擎系統3 0包括執行引擎8 0, 〇1&1〇§乂〇(11^顶語音處理單元300,3%八1^6(:(^111261^語音_別 器 302(可從 Speech Works® International Inc·獲得)、提示 單元304、記錄單元306,全部都在包括作業系統( 〇S)的服務邏輯執行環境(SLEE) 308中操作,且 是在硬體3 1 0上操作。S LEE是計算的環境,呼叫邏輯 在其中執行。此環境包括Artisoft® Visual Voice服務平台所 提供的任何工具,用以處理呼叫及事件。執行引擎8 〇被架 構成控制引擎系統3 0的操作。· 25 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) 474090 Α7 經濟部智慧財產局員工消費合作社印製 五、發明說明(23) 引擎系統3 〇被架構成識別語音並適應新語音。語音處 ϋ單元3 0 0被架構成處理所接收之呼叫者1 2發言及 S E L L記錄(下文中討論)的波形,並提供處理過的資料 給識別器3 0 2。識別器3 0 2使用儲存在ϊ V R系統1 6 Ψ的聲音模型、語意模型(字句的機率)、發音曲線圖以及 字典’嘗試識別呼叫者發言中的字或許多字。聲音模型表示 指定波形與語音中相關部分關連的統計機率。識別器3 0 2 可以產生最像呼叫者12所說之Ν個字或片語的Ν-最佳表 ° Ν -最佳表中的每一項目具有一對應的信賴分:數,用以代 表所識別之字或片語就是呼叫者1 2所說之語的可能性。引 擎系統3 0被架構成每當引擎系統3 0執行時即會載入控制 引擎系統3 0的模型與參數。這些資料儲存在組態與記錄系 統3 4中’如此,它們可在離線的批次處理模式中複製。 識別器3 0 2可建立离增加字典,並調適聲音模型、語 意模型、發音曲線圖以調整發言波形與語音間鏈結的槪率。 聲音模型的再訓練可以在I V R系統1 6不工作的期間執行 。識別器3 0 2被架構成使用解析及原文建立及評估語倉模 型,並使用S L Ε Ε記錄自動建立語意模型。 爲實施自動陪從功能,引擎系統3 0被架構成執行安排 呼叫線路的功能。這些功能包括識別員工的姓名及/或部門 及/或職稱,包括員工之姓名、部門及/或職稱的同義字, 並將提示提供給呼叫者。爲執行安排呼叫線路特性,執行引 擎8 0被架構成從組態與記錄系統3 4中擷取資訊,並將此 與呼叫者的語音比較,以決定與I V R系統1 6相關的公司_ (請先閱讀背面之注意事填寫本頁) 參 裝Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs 474090 A7 B7 V. Description of Invention (14) The station provides post tracking service or tariff calculation. The same service can be accessed via telephone using embodiments of the present invention, whether or not the service is provided on a website. The caller simply responds to the system's greeting and says the desired service name. The information provided in these services can be linked in real time to external content providers to provide information, as well as more complex actions. Information retrieval applications can include very simple information updates, such as tracking packages that have been delivered using a courier service, tracking airline luggage, obtaining bank account balances, and more: and more complex actions. Another application for information retrieval includes providing driving directions for callers who want to reach a location in the enterprise. The caller calls a voice system according to an embodiment of the present invention and responds to a greeting such as "company information 'direction" and the like. In response to this, the caller was prompted with "Where are you coming from?" The caller responded with a direction and identifiable points, such as the main road. In response to this ’the caller heard" I will read your direction ". The direction is then presented to the caller. The result is that speech provides useful information retrieval. The information retrieval function also includes retrieval of press releases, information sheets, and other electronic documents, as well as electronic text # transmission via text-to-speech, fax, or other media. .. Applications can use a variety of method architectures. For example, embodiments include tools, software packages, and framework wizards, which can be used by operators to create a new application that provides different information retrieval and transaction processing functions. Therefore, the disclosed embodiments can improve telephone response, leverage the company's investment in the World Wide Web, leverage leverage, and provide a variety of information retrieval and transaction processing features or network ports. The embodiment of the present invention provides interaction _ This paper size is applicable to the Chinese National Standard (CNS) A4 specification (210 X 297 mm) (Please read the notes on the back first to complete this page) «is Consumer Cooperatives, Intellectual Property Bureau, Ministry of Economic Affairs Printed -17- 474090 A7 B7 V. Description of the invention (15) Please read the notes on the back (please fill in this page). Voice system, organization, information content and services follow the network-based model. Users can use it. A telephone makes a call and accesses information and / or services in a natural language and interactive voice response (IVR) system. Therefore, embodiments of the present invention allow users to arrange lines to selected company employees via name and / or department, and provide organizations, nouns, and orders of similar websites to access company information and transactions. Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economics A typical embodiment of the present invention includes a basic platform and tool set, and a set of application modules that can be structured and pre-packaged. With the basic platform and toolset, you can use the toolset to customize a system to provide an individualized: customized interactive voice application. : With the collection of structured and pre-packaged application modules, basically customers buy a complete set of products, and only need to make a few modifications to configure the product shelf to meet customer needs. As the website-like function provided by the embodiment of the present invention in the I VR system, the embodiment of the present invention is called a SpeechSite ™ IVR system, and includes a SpeechSite ™ IVR interface. The speech pages in the SpeechSite ™ IVR system are similar to web pages, providing information and / or services, providing different groups or types of information and / or services with different speech pages, and the information and / or services usually provided by websites The service is the same and is provided by a typical website organization. The following description assumes that a company purchases and uses the described embodiment. Therefore, it is assumed that the described embodiment provides information and products / services related to the company purchased. However, entities different from the company are also acceptable. Structure of the whole system-18- This paper size is in accordance with Chinese National Standard (CNS) A4 specification (210 χ 297 public love) 474090 ..f A7 B7 Printed by the Employees' Cooperative of Intellectual Property Bureau of the Ministry of Economic Affairs 5. Description of invention (16) See _ι1: Interactive voice system 10 includes users and / or callers 12, Public Switched Telephone Network (PSTN) 14, IVR system 1 6. Simple Mail Transfer Protocol (STP) server 1 8. Firewall 2 0, network 22, here is the Internet, and analysis / report (A / R) service 2 4. As shown in the figure, the communication between each component in the system 10 is bidirectional. Caller 12 enters telephone 26 and fax machine 28. Caller 12 can communicate with PSTN 14 via telephone 26 or fax machine 28 and vice versa. The caller 12 communicates with the IV system 16 via the PSTN 1 4. I V R system 16: Prompt caller 12 in a guided conversation format to interact with caller 12 and recognize (or at least try to recognize) the voice from caller 12. The I V R system 16 also communicates with the A / R service 2 4 via the Internet 2 2. The SM server 18 provides an interface between the I VR system and the Internet 22. The firewall 20 uses conventional techniques to protect the communication between the I VR system 16 and the Internet 22. IVR system 16 includes engine system 30, management system 3 2, and configuration and recording system 3 4 0 system 30, 3 2, 3 4 handles the interaction between IVR system 16 and call luxury 12, which are used for The architecture engine system 30 stores configuration parameters, prompts, and other data, and records interactions with callers 12 as described in more detail below. Brief introduction of Baidong Voice Response System IVR system 16 can be implemented using a personal computer. For example, the following group _ This paper size applies to China National Standard (CNS) A4 specification (210 X 297 mm) -19- ----;- --------— Body --- (Please read the notes on the back to fill out this page first) Word · Γ 474 090 Α7 _-_____ Β7 5. Description of the invention (17) The items and / or features can be used as Part of a computer implementing IVR system 16: a single-processor workstation using I NTEL®PENT I UM® (please read the notes on the back first and fill out this page) III (NT workstation certification) processor, the timepiece rate 4 5 〇 Μ Η z or higher; 384Mb RAM or more; 9GB hard disk space; and high-speed DLT backup system; 10/100 Ethernet connection and 56K modem for connection; monitor; mouse; Keyboard for data display, input and operation; D41 ESC and D240 SC-T 1 telephone interface cards; Antares 6000/50 digital signal processor; NT4.0 workstation service pack 5 operating system; Artisoft® 5.0 Enterprise environment ; Access or SQL server; IIS or Pure Information Services HTT P server, and Microsoft® FTP Services for Windows NT Services using FTP Services, or Apache Software Foundation HTTP server; Lucent single line license for text-to-speech; PolyPM or PCAny where program for remote (E.g. desktop). Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs. Please refer to Figure 2. The computer system 5, 0 implementing the IVR system 16 includes a bus 5 2 or other communication mechanism for transmitting information, coupled with the bus 5 2 The processor 54 is used to process information. The computer system 50 also includes a main memory 56, such as a random access memory (RA M) or other dynamic storage device, coupled to the bus 5 2 for storing information and instructions for the processor 5 4 to execute. . During the execution of instructions by the processor 54, the main memory 56 is also used to store temporary variables or other intermediate information. The computer system 50 also includes a read-only memory (ROM) 58 or other static storage device coupled to the bus 52 to store static information and instructions used by the processor 54. Storage device 60, such as magnetic disks or optical disks, constitutes storage information and instructions, and is also coupled _ -20-This paper size applies to China National Standard (CNS) A4 (210 X 297 public love) 474090 A7 B7 Intellectual property of the Ministry of Economic Affairs Printed by the Bureau's Consumer Cooperatives V. Invention Description (18) Combined with the bus 5 2. The computer system 50 is coupled to a display 6 2 ′ such as a cathode ray tube (C R T) / to display information to a computer user via a bus 5 2. The input device 64 is coupled to the bus 5 2 such as a key sound of an alphanumeric key or other keys to transmit information and a selected command to the processor 54. Other types of user input devices in the system I. 5.0 include cursor control 66, such as a mouse, trackball, or cursor direction key J for transmitting direction information and selected commands to the processor 54 for Controls the movement of the cursor on the display 62. This input device typically controls the cursor on two axes, the first: an axis such as X, and the second axis such as y, allowing the device to indicate the position on a plane. According to an embodiment of the present invention, the computer system 50 response processor 54 executes one or more instructions contained in the main memory 56 one or more times, which can generate a speech recognition application. These instructions are read from other computer-readable media, such as the storage device 60 to the main memory 56. The processor 5 4 sequentially executes the instructions contained in the main memory 56 to cause the processor 54 to execute the method described herein. In other embodiments, hard-wired circuits, firmware, any kind of hardware combination, and / or software can be used to implement embodiments of the invention. ‘A computer-readable medium’ as used herein includes any medium that can provide instructions to a processor 54 for execution. There are many types of such media, including but not limited to non-volatile media, volatile media, and transmission media. Non-volatile media includes, for example, optical or magnetic disks, such as storage devices 60. Volatile media include dynamic memory such as main memory 56 '. Transmission media include coaxial cables, copper wires, and fiber optic cables, including the wires that make up the bus 5 2. Transmission media can also be ultrasonic or electromagnetic (such as light waves), such as during data communication (please read the notes on the back to fill out this page) • i Γ This paper size applies to China National Standard (CNS) A4 specifications (210 X 297 mm) -21-474090 Α7 -------_ B7 V. Description of the invention (19) The radio waves and infrared rays generated. .. The general types of computer-readable media include, for example, floppy disks, hard disks, magnetic tapes or other magnetic media, CD-ROM, any other optical media, hole cards, paper tapes, any other physical with hole patterns Media, RAM, PROM, EPROM, FLASH — EPROM 'Any other memory chip or cassette (such as electrical and / or electromagnetic, including optical), a carrier wave as described below, or any other computer-readable media. Various types of computer-readable media can carry one or more sequences of one or more instructions for execution by the processor 54. For example, at first, the instructions could be on a disk of a remote computer. The remote computer can download the instructions to its dynamic memory and use a modem to transmit the instructions over a telephone line. The modem located at computer system 50 can receive data from the telephone line, and use an infrared transmitter to convert the data into an infrared signal. The infrared detector can receive the data contained in the infrared signal, and there is an appropriate circuit to The data is put on the confluence and row 5 2. The bus 5 2 can load data to the main memory 56, and the processor 5 4 can fetch instructions from the main memory 56 and execute them. The instructions received by the main memory 56 may be selectively stored in the storage device 60 before or after the processor 54 executes. The computer system 50 also includes a communication interface 6 8 coupled to the bus 5 2. The communication interface 68 provides two-way communication, which is coupled to the network link 70 and then coupled to the SM TP server 18. For example, the communication interface 68 can be an integrated service digital network (ISD Ν) card or a modem. , Connect to the corresponding type of telephone line to provide data communication. In another example, the communication interface 6 8 may be the paper size applicable to the Chinese National Standard (CNS) A4 specification (210 X 297 mm) (please read the notes on the back first to complete this page). Words Γ Wisdom of the Ministry of Economy Printed by the Consumer Cooperative of the Property Bureau-22- 474090 A7 Printed by the Consumer Cooperative of the Smart Property Bureau of the Ministry of Economy v Wireless links can also be used. The communication interface 68 can transmit gas and / or electromagnetic (including optical) signals. These signals carry digital data streams of type information. The computer system 50 can send messages and receive data through the SMTP server 18, 70, and the communication interface 68, including Cheng Ru. You can download the code from the Internet 2 2 (Figure 1) for medium The described speech recognition application. The processor 54 can receive the code on the receiving line and / or store it in the storage device 60 or other sex memory for later execution. According to this method, the computer system obtains the application code in the carrier type. Now referring to FIG. 3, the IVR system 16 includes an engine system management system 3 2, a configuration and recording system 3 4, a remote control system (R 36, a support system 38, and a monitoring interface system 40. As shown by arrows in FIG. 3 These systems can communicate in both directions. In addition, the remote control system is responsible for the two-way communication systems 32, 34, 34, 38, and 40. 2 2 is responsible for the architecture, reporting, and monitoring of the IVR system 16. 2 Includes network application servers 4 2 and the application server logic 4 Management system 3 2 To be applied via the network application server 4 2, such as servers, of course, other servers can also be accepted. The application server is controlled by the application server logic 4 4, which is It is implemented in software. The communication link and receiving power represent a variety of network link codes. For example, the non-volatile 50 can 30 and the two-way system 3 and 6 in the non-volatile 50 and 30 can be executed when the code is generated. 4. Department of Management Access to ΤΤΡ service 4 2 ticket management system management system 3 2 is responsible for the construction of other components of the I VR system 16 ----------------- (please first Read the notes on the back and fill in this page) Order · 474030 A7 ___B7 V. Description of the invention (21) The management system 3 2 is constructed to access the information stored in the configuration and recording system 3 4: Use this information as a configuration Information is provided to other components of the IVR system 16. For example, the management system 3 2 is configured to read configuration data from the configuration and recording system 3 4 and provide this information to the engine system 3 0. The configuration data transmitted to the engine system 30 includes data used to determine which voice page is to be activated, and the page content includes the location of the prompt, which grammar to use, and which vocabulary to use. The voice pages are grouped into the voice module according to the configuration information of the management system 32. Different voice modules provide different types of information and services:. Each module contains many voice pages. Typical examples of modules are Speech Attendant, Contact Us, and Company Information. The voice companion module includes a page of people directories in the company. The Contact Us module includes information for contacting the company, such as email addresses, mail addresses, street names and addresses, fax numbers, and telephone numbers. The company information module includes pages describing general information about the company, such as the type of business and / or services provided by the business, news, and the geographic area of the service. The management system 32 is also structured to update / edit the information contained in the configuration and recording system 34. Therefore, for example, the management system 32 can access and edit employee name and phone number lists, company information such as news, and call flow control data. Engine system. The engine system 30 is responsible for the construction of the voice interface between the IVR system 16 and the caller 12 (Figure 1), and the connection to the telephone system. This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) (please read the note on the back first to fill in this page) lii Printed by the Intellectual Property Bureau of the Ministry of Economic Affairs Consumer Cooperatives -24- 474090 A7 B7 V. Description of the invention (22) (Please read the note on the back first to fill in this page) Therefore, the engine system 30 is connected to PSTN 1 4 for two-way communication, and it is suitable to perform the call flow with caller 12, including identification Voice, play prompts, retrieve data, schedule call lines, disambiguate responses, confirm responses, select pages to link to, and connect to supporting devices (such as TTS and fax). The engine system 30 is constructed using conventional techniques to recognize speech, analyze the speech into small segments, and apply a sound model. The engine system 30 functions were implemented using D ^ logModule ™ speech processing software from Speech Works® International Inc. of Boston, Mass. Arrange call lines, retrieve information, and process transactions. Each function is the speech engine 30 trying to recognize: not the caller's voice, and the actions taken accordingly. The engine system 30 also includes an execution engine 80, which is framed to form a process that controls the engine system 30. This process includes capturing and playing prompts, speech recognition, and monitoring and reporting engine system 30 and caller 12 interactions. The engine system 30 is stored in the command and / or data control provided by the engine system 30 or the configuration and recording system 34 or the caller 12. Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs, please refer to FIG. 10, the engine system 30 includes the execution engine 80, 〇1 & 1〇§ 乂 〇 (11 ^ top voice processing unit 300, 3% eight 1 ^ 6 (: (^ 111261 ^ Voice_Baker 302 (available from Speech Works® International Inc ·), prompt unit 304, recording unit 306, all in the service logic execution environment (SLEE) including the operating system (OS) It operates in 308 and operates on hardware 3 10. S LEE is a computing environment in which call logic is executed. This environment includes any tools provided by the Artisoft® Visual Voice service platform to handle calls and events. The execution engine 8 〇The quilt constitutes the operation of the control engine system 30. · 25 This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) 474090 Α7 Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs Description of the Invention (23) The engine system 3 〇 frame constitutes recognition speech and adapts to the new voice. The voice processing unit 3 0 0 frame constitutes processing the received caller 12 speech and SELL record (discussed below) waveforms And provide the processed data to the recognizer 3 0 2. The recognizer 3 0 2 uses the sound model, semantic model (probability of words), the pronunciation curve graph and the dictionary 'try to recognize the caller's speech stored in the ϊ VR system 1 6 Ψ. Words or many words. The sound model represents the statistical probability that the specified waveform is related to the relevant part of the speech. The recognizer 3 0 2 can produce an N-best table of N words or phrases that are most like what the caller 12 says Ν-Each item in the best table has a corresponding trust score: number, which represents the possibility that the recognized word or phrase is the language spoken by the caller 12. The engine system 30 is framed to form each When the engine system 30 is executed, the model and parameters of the control engine system 30 are loaded. These data are stored in the configuration and recording system 34. Thus, they can be copied in offline batch processing mode. Recognizer 3 0 2 can be built to increase the dictionary, and adjust the sound model, semantic model, pronunciation curve graph to adjust the probability of the link between the speech waveform and speech. Retraining of the sound model can be performed during the IVR system 16 is not working. knowledge The device 302 was constructed using the analysis and original text to build and evaluate the language warehouse model, and the SL Ε Ε record was used to automatically build the semantic model. In order to implement the auto-accompaniment function, the engine system 30 was constructed to perform the function of arranging call lines. These functions include identifying the employee ’s name and / or department and / or job title, including synonyms of the employee ’s name, department, and / or job title, and providing prompts to callers. To schedule call line features for execution, the execution engine 80 is The frame structure extracts information from the configuration and recording system 34, and compares this with the caller's voice to determine the company related to the IVR system 16_ (Please read the precautions on the back and fill out this page first)
A 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公爱) -26- 474090 A7 B7 五、發明說明(24) 或事業中是否有人或部門與所說的姓名、部門或職稱對應。 引擎系統3 0可以消除回應的岐義,例如,提示呼叫者1 2 ,除了員工姓名之外 '還要確認員工所在的部門。安排呼叫 線路的功能也包括執行引擎8 0將呼叫者1 2聯絡到所ΐ靑求 的人/部門。特別是,執行引擎8 0被架構成使用blind fl a s h - h 〇 ◦ k傳送,按照組態與記錄系統3 4中所儲存的資料 傳送呼叫。引擎系統3 0也被架構成執行其它類型的傳送, 例如受監督的傳送。 對資訊擷取功能而言,引擎系統3 0可以識別呼叫者 1 2所請求的指定語音頁,或決定那一頁中包含所請求的資 訊。引擎系統3 0被架構成識別來自呼叫者1 2的語音,以 便決定呼叫者1 2請求何種資訊。回應所識別之呼叫者1 2 的語音,引擎系統3 0被架構成存取指定/決定的頁,並播 放與請求之資訊相關的提示給呼叫者。因此,當引擎系統 3 0在適當的頁時(提供與”與我們聯絡”頁鏈結的頁、), 例如可以回應使用者/呼叫者1 2所說的”與我們聯絡”而 鏈結到”與我們聯絡”頁.。此外,當引擎系統3 0在適f的 頁時(提供與”與我們聯絡”頁鏈結的頁),例如可以回應. 使用者請求”與我們聯絡”頁中的資訊而鏈結到”與我們聯 絡”頁。例如,呼叫者1 2說”到波士頓的方向”,引擎系 統3 0即將呼叫者1 2鏈結到”與我們聯絡”頁。 引擎系統3 0也被架構成呼叫者1 2可經由特定的 I V R系統1 6執行交易。爲執行交易,引擎系統3 0被架 構成識別一指定的頁或決定包含呼叫者1 2所請求之資訊或- 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公爱) ----&---·-------裝--- (請先閱讀背面之注意事0填寫本頁) · 經濟部智慧財產局員工消費合作社印製 -27- 474090 A7 __ B7 五、發明說明(25) 服務(例如股票交易)的頁’並存取指定/決定的頁,播放 提示給呼叫者1 2,並執行/起始呼叫者1 2所指定的功能 。引擎系統3 0可識別呼叫者的語音,並將被識別的語音與 儲存在組態與記錄系統3 4中的資料關聯。按照儲存在引擎 系統3 0中的指令及/或儲存在組態與記錄系統3 4中的資 料(包括適當的指令),引擎系統3 0可以按照呼叫者1 2 所提供的資料,執行所指示的交易。 爲控制與呼叫者1 2的對話,引擎系統3 0經由 PSTN 1 4及組態與記錄系統3 4與呼叫者:1 2互動。 引擎系統3 0可以經由P S T N 1 4接收呼叫者1 2的語 音。爲識別語音,引擎系統3 0可以在組態與記錄系統3 4 中存取用以指示何種語音可被I V R系統1 6識別的資訊。 引擎系統3 0被架構成在執行引擎8 0的控制下,管理 I V R系統1 6與呼叫者1_ 2間的交談。視引擎系統3 0是 否識別了呼叫者1 2的語音而定,執行引擎可以指令引、擎系 統3 0輸出儲存在組態與記錄系統3 4中的提示給呼叫者 1 2。這些提示可以是,例如,依照先前識別的語音,#叫 者1 2所請求的資訊,請呼叫者1 2重新說出無法識別的語 音或被識別但信賴度低的語音,或對引擎系統3 0所接收到 之非語音資訊發出其它適當的錯誤訊息。 引擎系統3 0被架構成以指引對話的方式與呼叫者1 2 聯絡,以引導呼叫者1 2以有限的字彙獲得呼叫者想要的結 果。引擎系統3 0呈現給呼叫者可以說出的命令,以及使用 認可的字彙,其中包括可能的命令(例如”與我們聯絡”) 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) (請先閱讀背面之注意事填寫本頁) --裝 經濟部智慧財產局員工消費合作社印製 -28- 474030 A7 B7 五、發明說明(26) ,以及一些同義字或其它傳送相同意義的字(例如”方向” )。認可的字彙隨著與呼叫者1 2間互動對話狀態的不同而 改變。呼叫者1 2也可說出適用於任何語音頁之”全面的” 或”通用的”命令,例如”說明”、”上一頁”、”下一頁 ”。”上一頁”與”下一頁”命令只有在目前的語音頁在呼 叫者12曾經進入之語音頁中的前或後方仍有語音頁才適用 。在允許之外的發言’呼叫者1 2都會得到錯誤訊息。限制 可用的可識別語音有助於增進識別的精確度及速度,以及 I V R系統1 6整體的穩定度。 : 引擎系統3 0被架構成使用習知的技術識別及回應呼叫 者1 2的請求。例如,來自呼叫者1 2的語音可以被解析成 語音單元,並由數位信號處理器轉換產生語音單元的向量。 這些向量被群組成長度不同的語音段。這些段再被轉換成特 徵向量,這些特徵向量再被語言限制(例如認可的字彙)分 析,產生具有最高信賴分數之N個字串的N -最佳表。 引擎系統3 0與呼叫者1 2間的互動是經由使用者介面 呈現,它將典型網站上所提供的視覺格式轉換成聲音格式, 包括網站瀏覽器所提供的功能。使用者介面由引擎系統3 0 所存取的提示完成,並呈現給呼叫者1 2。引擎系統3 0播 放的提示儲存在組態與記錄系統3 4中,與提供給瀏覽網站 之人相同的資訊,例如告知呼叫者目前的位置(例如”首頁 ”)或所要移動到的位置(例如”移動到與我們聯絡頁”) 。這些播放給使用者的提示使用與網站相關的術語,例如” 頁”、”與我們聯絡”、”公司資訊”、”回到第· ··頁 (請先閱讀背面之注意事ί填寫本頁) 裝 i' 經濟部智慧財產局員工消費合作社印製 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) -29- 474090 A7 B7 五、發明說明(27) ”、”前進到第· ·,頁”。在其它的例中,播放給呼叫者 12的提示可以是”你可說〔第1頁〕〔第2頁〕··· · (請先閱讀背面之注意事HI填寫本頁) ”。〔第1頁〕〔第2頁〕的提示是可架構的文字提示,例 如是儲存在組態與記錄系統3 4中的客戶文字,並被擷取播 放給呼叫者1 2。這些文字取代致候語中的字,例如,引擎 系統3 0根據儲存在組態與記錄系統3 4中的資訊將客戶文 字即時提供給呼叫者1 2。典型的客戶頁例如公司的特定產 品,公司的特定部門及/或公司的特定服務。還可以呈現其 它詢問的提示給呼叫者1 2,例如詢問是/否。引擎系統 3 0呈現給呼叫者1 2之資訊的組織可與對應之網站的內容 不同,但基本上呈現給呼叫者1 2相同的資訊。 如同類-網站使用者介面的一部分,引擎系統3 0被架 構成回應類-網站的命令導覽IVR系統16中的整個資訊 。呼叫者可以說出網路瀏覽器所提供的典型命令,例如”首 頁”、”上一頁”、”下一頁”、”說明”及”到”、引擎 系統3 0被架構成認可這些命令並採取動作。因此,引擎系 統3 0回應呼叫者1 2所說的上述命令,在曾經進入的頁:中 ,將呼叫者1 2從所在頁移動到前一頁、後一頁或回到首頁 經濟部智慧財產局員工消費合作社印製 〇 對每一語音頁而言,它們可鏈結到引擎系統3 0提示給 呼叫者1 2的特定頁。例如,首頁可以具有到公司資訊頁、 與我們聯絡頁、以及產品/服務頁的特殊鏈結,並將鏈結提 示給呼叫者1 2。例如呼叫者1 2可以被告知”您可說”公 司資訊”、”與我們聯絡”、或”產品及服務””。引擎系_ -30- 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) 474090 A7 --- B7 五、發明說明(28) 統3 0可以按呼叫者1 2的請求移動到這些特定的頁。、 -------*-------裝--- (請先閱讀背面之注意事β填寫本頁) 引擎系統3 0也被架構成提供蒐尋服務並能自動將服務 傳真給呼叫者1 2。在與I v R系統1 6對話的適當點,呼 叫者1 2可以說”尋找”。回應此請求/發言,引擎系統 3 0在所儲存的語音頁中蒐尋所指示的文字及/或資訊。呼 叫者1 2可以在任何時間說出”傳真它”,且引擎系統3〇 被架構成回應此請求,將目前頁的內容(與目前語音頁有關 的內谷)傳真到呼叫者1 2所指定的傳真號碼。傳真號碼可 事件先儲存並確認,或引擎系統3 0在回應傳真命令後請求 傳真號碼。 經濟部智慧財產局員工消費合作社印製 引擎系統3 0被架構成將呼叫事件以及其它的交易記錄 到組態與記錄系統3 4中。呼叫事件是I V R系統1 6與呼 叫者1 2間互動交談的各階段。這些事件包括呼叫者1 2的 請求,引擎系統3 0所做的識別,指示呼叫者的語音是否被 識別,語音是否因識別信賴度低或未被識別而不被接受、,引 擎系統3 0所起始的動作,以及播放提示給呼叫者1 2。事 件可以包括呼叫者想要到的頁,呼叫者1 2請求的命令/是 以何順序請求,以及引擎系統3 0所執行的動作。引擎系統 3 0被架構成將呼叫事件的指示傳送到組態與記錄系統3 4 儲存供未來參考。引擎系統3 0被架構成每當呼叫事件發生 時送出某些呼叫事件的指示.,同時,只有在發生某些情況時 才傳送其它呼叫事件的指示,以及,可被架構成完全不送出 其它呼叫事件的指示。例如引擎系統3 0可以在每次發生信 賴度低不接受時,或信賴度高而接受時送出指示(例如,某 -31 - 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) 474030 A7 ___ B7 五、發明說明(29) (請先閱讀背面之注意事填寫本頁) 人成功地連接使用自動陪從特性 >。引擎系統3 〇也被架構 成根據呼叫事件的呼叫統計產生報告,供儲存到組態與記錄 系統3 4中,及/或供監視介面系統4 0擷取。 組態與記錄系統 組態與記錄系統3 4包括記錄儲存區8 6、資料庫儲存 區8 8,以及一般儲存區9 0。組態與記錄系統3 4被架構 成儲存管理系統3 2、引擎系統3 0、支援系統3 8以及監 視介面系統4 0所使用的資訊,且與這些系統每一個都是雙 向互動。因此,這些系統30、32、38、40中的每一 個都可從組態與記錄系統3 4擷取資訊,或將資訊儲存到其 中0 經濟部智慧財產局員工消費合作社印製 資料庫8 8儲存組態檔及語音頁的內容。語音頁的內容 與一般從網站上獲得的內容及格式相似。組態檔是用來架構 系統30、32、36、38、40。在架構及建立期、間, 架構這些系統30、32、36、38、40每一個所需的 資訊儲存在這些組態檔中。這些檔案可由I V R系統1弓的 製造商及/或購買者建立及/或修改,以提供及/或改變客 戶的組態。資料庫8 8也被架構成儲存與語音頁有關的各種 資訊。例如,資料庫8 8被架構成儲存與提示相關的資訊。 提示資料包括識別、錄製日期、錄製提示的人名、來源、以 及提示的類型。此外,提示是否被公開、用於提示之特有使 用者介面名稱、以及提示的文字都儲存在資料庫8 8中。此 外,提示在資料庫8 8中的位置以及產生提示的日期也儲存_ -32- 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) 474030 A7 B7 經濟部智慧財產局員工消費合作社印製 五、發明說明(3〇) 在資料庫8 8中。將多個提示鏈結在一起以構成片語或其它 提示段的資訊也儲存在資料庫8 8中。1 資料庫8 8也儲存與語音模組相關的資訊。此資訊包括 …........ ... _________________________________________ 語音模組與語音頁的識別資訊,包括每一個模組及頁的內容 。此識別資訊適合供引擎系統3 0用來定位、擷取及處理各 種語音模組及語音頁,提示也包括在其內。 資料庫8 8也儲存與語音頁相關的資料。頁與頁之組成 間的鏈結包含在資料庫內,且有助於引擎系統3 0鏈結到其 它頁及/或模組,以更容易地擷取資訊及執行其:它動作。資 料庫8 8也儲存DialogModule™語音處理單元3 0 0 (圖 1 〇 )與特定語音頁及/或語音模組鏈結的資訊。 〇^1〇£乂(^111^巧吾音處理單元3 0 0 (圖10)的鏈結資訊 提供一映射,用以決定當執行語音頁時要執行那一個 DialogModule™語音處理單元3 0 0 (圖1 0 )。 儲存在資料庫8 8中的資料也提供做爲同義字之資、料間 的相互鏈結。當某項目有同義字可用時,有助於增加識別該 項目的精確度。 ' 資料庫8 8也儲存其它幾種類型的資訊。此資訊包括有 助於支援語音頁導覽的資訊,包括鏈結到導覽功能的導覽名 詞,以及當引擎系統3 0中的執行引擎8 0執行”尋找”功 能時用來定位語音頁的關鍵字。使用者字典也儲存在資料庫 8 8中。資料庫8 8也包含與公司營運相關的資訊。例如公 司營業的日期及/或時間也儲存在資料庫8 8中。資料庫 8 8也儲存個人目錄的資訊,以提供自動陪從功能。儲存在 (請先閱讀背面之注意事填寫本頁) 11¾ ;裝 丨線」 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) -33- 474090 A7 B7 五、發明說明(31) (請先閱讀背面之注意事Ιϋ:填寫本頁) 資料庫8 8中之個人目錄的資訊是儲存在資料欄中。攔位例 如包括人員姓名、暱稱、職稱、部門、這些攔位之登錄的同 義字,以及個人、辦公室、部門的電話分機,或傳送/安排 呼叫線路到個人及/或部門等的其它資訊。這些欄位可以更 新以反映新進人員、部門的改變、部門名稱的改變、名稱及 人員的改變以及增加暱稱或其它同義字。 、' r 所儲存之語音頁的內容包含IVR系統16之每一頁所 有的資訊,包括提示(例如詢問及資訊)、配置及1結1 >語 #頁的內容被劃分成可以被選擇的資料欄,以便‘在送給購買 者/客戶前,將頁的內容修改成客戶所要的架構,或由 I V R系統1 6的客戶自行修改。語音頁的內容可按需要修 改資料欄而更新,例如,更新股票價格,提供每日新聞,或 指出公司可能發生的任何改變。 經濟部智慧財產局員工消費合作社印製 儲存區9 0儲存所有的提示、傳真頁、G I F s、以及 語音模型。提示全部爲聲音資訊,用以轉播給呼叫者2。 例如,提示包括詢問呼叫者1 2以及傳送給呼叫者1 2的資 訊陳述。傳真頁是回應呼叫者傳真資訊的請求(例如呼til[者 1 2說”傳真它”等)傳送到呼叫者1 2之傳真機2 8 (圖 1 )的資料。圖形資訊是圖形互換格式(G I F )之結構的 檔案,可包括在傳真頁內。語音模型供引擎系統3 0用來識 別部分語音,以識別呼叫者1 2所說的字及/或片語。 記錄儲存區8 6被架構成儲存呼叫事件之記錄及系統 4 〇所需的其它資訊,例如服務邏輯及執行環境(S L E E )記錄。呼叫事件的記錄包括相關的統計’例如,呼叫的時 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) 474090 A7 B7 五、發明說明(32) 間、呼叫的長度r請求的語音頁I、識別的成功率、認可之外 的字彙、失敗的識別,;以及使用的命令。 支援系統 支援系統3 8被架構成可被管理系統3 2及/或引擎系 統3 0請求,並提供這些系統3 0、3 2所需的支援。支援 系統3 8包括文字一到一語音(TTS)特徵9 2、記錄轉 換器9 4 '傳真特徵9 6、、報告產生器9 8及語音調適器 1〇〇。 T T S 9 2允許引擎系統3 0根據儲存在I V R系統 1 6中的文字輸出語音或其它適當的聲音給呼叫者1 2。實 施T T S 9 2可以使用習知的技術,如Lucent T T S弓[ 擎。T T S 9 2允許I V R系統1 6很快速地更新。例如 ,新聞稿可以很快速地以文字的形式儲存到組態與記錄系統 3 4中,使用T T S 9 2及引擎系統3 0可立刻以語.音輸 出給呼叫者1 2。例如稍後由名新聞主播錄製成錄音,可以 用來取代T T S 9 2將文字新聞轉換成引擎系統3 0適用 的語音。I V R系統1 6的其它部分也可按此方式更新,例. 如,在人員目錄中的員工冊。 記錄轉換器9 4被架構成將儲存在儲存區8 6中之記錄 的資訊轉換成適當的格式供報告產生器9 8處理。在此,記 錄轉換器9 4被架構成存取儲存在儲存區8 6中的S L E E 檔案,並將這些檔案轉換成NC SA ( National Center for Supercomputing Applications)標準記錄。記錄轉換器 9 4 可- 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) (請先閱讀背面之注意事填寫本頁) 裝 · 經濟部智慧財產局員工消費合作社印製 -35- 474090 A7 _ B7 五、發明說明(33) 有效地將呼叫者對I V R系統1 6的存取指示轉換成等於網 頁的”尋到” (hit),並將這些尋到儲存到檔案中。因 此,記錄轉換器9 4被架構成儲存一檔案,其中包含呼叫者 1 2之I D (例如電話號碼)、呼叫者1 2請求的日期及時 間,以及呼叫者1 2對資訊或動作之請求的指示。由記錄轉 換器9 4所儲存的記錄被儲存到組態與記錄系統3 4中。 傳真特徵9 6被架構成處理來自呼叫者1 2的傳真請求 ,並將所請求的資訊傳真到呼叫者1 2可取得的傳真機2 8 (圖1 )。例如,傳真特徵9 6可以使用WmFax Pro 9.0實 施。傳真特徵96的此種實施支援RightFax伺服器及網際網路 伺服器。傳真特徵9 6可以將資訊傳真到與傳真機2 8相關 的傳真號碼,傳真號碼是呼叫者1 2經由傳真伺服器9 7提 供。 報告產生器9 8被架構成存取儲存在組態與記錄系統 3 4的記錄及其它資訊,並操作這些資料以產生各種報、告。 例如,報告產生器9 8可以操作記錄轉換器9 4所儲存的記 錄,以產生與被尋到之語音頁相關的報告。報告產生器自8 被架構成產生指示每小時呼叫次數的報告,所有語音模組每 小時呼叫次數的報告,以及每小時操作員傳送次數的報告。 報告產生器9 8也可以產生在所選擇的日/週/月,由自動 次數識別器(A N I )所識別來自某指定裝置之呼叫次數的 報告。所產生的這些報告,可以是撰寫及繪圖格式,且可下 載及匯入到資料庫。 語音調適器1 0 0被架構成供引擎系統3、0使用的調適_ --------*-------裝--- (請先閱讀背面之注意事'ΐϋ填寫本頁) IT· 經濟部智慧財產局員工消費合作社印製 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) -36- 474090 A7 B7 經濟部智慧財產局員工消費合作社印製 五、發明說明(34) 工具’以有助於增進引·擎系統3 0的語音識別。語音調適器 1 0 0 可以使用麻州波士頓之Speech Works® Inter*riati〇nal Inc.的LEARN 6 · 0軟體實施、。語音調適器1 〇 〇可以存取 儲存在組態與記錄系統3 4中的資訊,分析此資訊,並決定 如何調適儲存在組態與記錄系統3 4中的聲音模型、發音曲 線圖及/或語意模型。語音調適器1 0 0也被架構成按照所 做的決定更新/改變聲音模型、發音曲線圖及/或語意模型 。新的模型及圖形再被儲存到組態與記錄系統3 4中,供引 擎系統3 0用來識別來自呼叫者1 2的語音。: 遙控系統 遙控系統(R C S ) 3 6被架構成經由類比通信線路 104提供IVR 16遙控。RCS 36包括遠端存取 系統(RAS) 106,由適當的軟體控制,在本文中是PC Anywhere 108。RAS 1 0 6經由數據機1 1 0與親比線 路1 0 4通信。 R C S 3 6容許經由N T視窗對I V R 1 6做住何 的控制。例如,R C S 3 6可被架構成容許開始/停止處 理,以修改系統3 0、3 2、3 8、4 0的組態,包括儲存 在其內的資料,存取管理系統3 2,以及起動/關閉連接到 I V R 1 6的通信線路。 監視介面系統 監視介面系統4 0提供對I V R 1 6的監視功能,包. ----.---.-------裝--- (請先閱讀背面之注意事_填寫本頁) · 線- 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) -37- 經濟部智慧財產局員工消費合作社印製 474030 A7 ________ B7 五、發明說明(35) 括系統監視器1 1 2、提示監視器1 1 4及調整監視器 1 16。這些監視器1 1 2、1 14及1 16的每一個都被 架構成以u 1 aw檔案(μ— 1 aw檔案’文字波形)的型 式從組態與記錄系統3 4擷取資訊以及將資訊儲存入其中’ 並與SMTP伺服器1 8雙向通信。提示監視器1 1 4被架 構成監視提示的改變,並當改變時提供警報。 系統監視器1 1 2被架構成監視1 v R 1 6的電腦功-會I,回應監視功能採取適當動作,並提供一”基本心博動” (base heartbeat)給A/R服務2 4 (圖1 )。:基本心博動 是送給A / R服務2 4的訊息,告訴A / R服務2 4 I V R 1 6是以正常的操作參數在操作及動作。系統監視 器1 1 2提供警報及警告給硬體,以及電話錯誤、資源限制 、執行時間錯誤、及/或交易的錯誤。資源限制的錯誤施加 到I V R 1 6中之應用軟體的碼。執行時間錯誤提供給 S L E E、語音識別器、及DialogModule™語音處理單、元程 式庫。S L E E程式庫被架構成接受來自呼叫者1 2的呼叫 ’並請求引擎系統30,包括它的語音識別器。IVR ‘ 1 6軟體碼中執行時間及交易的錯誤包括處理來自呼叫者 1 2之呼叫所遭遇的所有各種錯誤。系統監視器1 1 2也報 告交易錯誤,並將這些交易錯誤的指示.儲存到組態與記錄系 統3 4中。系統監視器1 1 2也被架構成執行某些補救動作 ,如重新開始選擇I V R 1 6之非關鍵的服務。系統監視 器1 1 2送出的警報及警告經由網際網路(圖1 )送給a / R服務2 4 (圖1 )。 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) -38- (請先閱讀背面之注意事?^填寫本頁) »1 474090 A7 B7 五、發明說明(36) (請先閱讀背面之注意事填寫本頁) ^整監視器1 1 6 §架構成監視及分析有關呼叫者1 2 與I V R 1 6間互動的語音性能。調整監視器1 1 6被架 構成從儲存在組態與記錄系統3 4的S L E E記錄中計算性 能統計,並追蹤性能統計。從性能統計中,調整監視器 1 1 6可以送出關於這些性能統計的警告。調整監視器 1 1 6也送出S L E’ E記錄及來自呼叫者1 2之語音中被標 示爲可能有問題之部分的波形供外部監視。調整監視器 1 1 6也可輸出有關於交談統計的狀態訊息。調整監視器 1 1 6將這些警告、記錄、波形及訊息在網際網路2 2 (圖 1 )上傳送給A / R服務2 4 (圖1 )。 經濟部智慧財產局員工消費合作社印製 調整監視器1 1 6被架構成提供許多與呼叫者1 2及 I V R 1 6間交談之性能統計有關的報告。調整監視器 1 1 6可以按照數項標準分析性能統計,如重要交易的交易 完成率,DialogModule™語音處理單元完成率、失敗的呼叫 、呼叫者感覺的回應時間、短呼叫的比例、呼叫之切斷、相對 傳送的次數、被傳送給客服員的呼叫次數,以及呼叫的音量 。那一筆交易該指示爲重要交易是在系統架設或修改後藥構 。DialogModule™語音處理單元完成率的資訊包括發生多少 次確認,以及發生多少次失敗。有關於DialogModule™語音 處理單元完成率的資訊,是按照與DialogModule™語音處理單 元3 0 0 (圖1 0 )相關的語音頁格式化。呼叫者感覺的回 應時間可以用來決定I V R 1 6的工作是否過荷。人員姓 名未被使用的時間預先決定,視需要選擇,例如1、6及/ 或1 2週。關於呼叫之切斷相對傳送的次數以及傳送給客服_ -39- 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公爱) 474030 A7 B7 五、發明說明(37) 員之呼叫次數的報告,對分析自動陪從性能十分有用。 (請先閱讀背面之注意事填寫本頁) 調整監視器1 1 6也提供數項業務報告。例如提供每小 時呼叫次數的報告,每小時重要對話段落之呼叫次數的報告 ,每小時客服員傳送的次數,以及在預定時段內,來自某 AN I的呼叫次數。所提供之每小時的呼叫次數是可下載的 格式,如文字與圖形·。重要對話段落是由儲存在組態與記錄 系統3 4中的組態檔定義。A N I報告的既定時間範圍例如 曰、週及或月。這些報告是以可匯出的格式提供,經由一文 字檔或載入資料庫供做商業資料及其它的報告功能。 調整監視器1 1 6可以在各種失敗及/或性能條件下觸 發警報。這些警報以結構訊息送出,使用例如單網路管理協 定(S NMP )或電子郵件送往一或多個目的地。警報有助 於經由客戶營運中心,離站外部資源監視公司,及/或銷售 實體對系統性能做離站(off-she)監視及/或架構I V R 16。 、 分析/報告服務 . 經濟部智慧財產局員工消費合作社印製 調整監視器1 1 6可透過網際網路2 2 (圖1 )將報告 送給A / R服務2 4 (圖1 )。現請再參閱圖1 ,A / R服 務2 4被架構成監視I V R 1 6的性能,並提供警告以起 始關於I V R性能的診斷動作。診斷動作例如可由I v r 1 6 的供應商採行,諸如Speech Works® International Inc. 。A / R服務2 4被架構成存取組態與記錄系統3 4內的資 料,分析這些資料,並決定適當的動作,產生適當的警告, -40- 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) 474090 A7 — B7 五、發明說明(38) 及/或產生適當的報告〜A / R服務2 4被架構成周期性地 存取及/或接收資料,諸如儲存在組態與記錄系統3 4中的 發言記錄及S L E E記錄的檔案,用以識別、調整、監視並 產生報告。 ' A / R服務2 4所能產生的一種警報是潛在的高〇〇V 率。會造成〇〇V率高的原因例如沒有維護儲存在組態與記 錄系統3 4之人員目錄中的姓名表。因此,當呼叫者1 2要 求某特定姓名時,,1 VR 1 6不接受所請求的姓名,儘管 所請求的人員是I V R 1 6所服務之公司的員工。 A / R服務2 4可以爲可能的候選同義字/暱稱識別產 生警告及/或報告。當未被識別的片語或信賴度低的片語在 再次嘗試時因識別(具高信賴度)而被接受(例如,呼叫者 :”貴公司的〇E C ” ; I V R : ”不瞭解,請說姓名”; 呼叫者:” Stuart Patterson/’ ),呼叫者12在第一次嘗試 時使用的片語,是第二次嘗試所識別之人員很好的候選、同義 字。A / R服務2 4可以產生一報告指示可能的同義字(例 如C E〇)及被識別的語音(例如Stuart Patterson)。 A/R服務2 4也可爲重複的不良發音產生警告。對呼 叫者1 2所說的某個片語認可的百分比高,表示I VR 1 6的規劃對該片語有關的發音差。指示識別字/片語之重 複認可的警告可用來起始一動作,以調整I V R 1 6的發 音識別,以減少某字/片語所需的認可次數。 A/R服務2 4也被架構成爲沒有被I VR 1 6識別 的名稱產生警報、警告及/或報告。先前信賴分數高的名稱一 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公嫠) (請先閱讀背面之注意· 事 填寫本頁) 經濟部智慧財產局員工消費合作社印製 -41 - 474030 Α7 _ Β7 五、發明說明(39) 現在的信賴分數低,指示出數項原因,如發音識別差及/或 雜訊位準高及/或該人是該公司的前員工,原表列於資料庫 8 8中,但現已離職。 A / R服務2 4被架構成監視信賴分布以有助於管理信 賴分數的認可門檻。I V R 1 6也可調適成增進呼叫者的 語音識別,以增加識別的準確度。按此做,可位移所有發言 之信賴分數的整個分布。不過,不接受之發言的百分比下降 也許指示僞接受上升(即,未被識別的也被接受),這是由 於接收的門檻太低。反之,如果不接受的門檻太:高,不接受 率被人爲地拉高,抑制了真正的識別準確度。A / R服務 2 4可以監視信賴分布並影響門檻以有助於獲得適當的門檻 ,以有助於實現真正的識別準確度。A / R也可以產生個人 信賴分數及不接受率的警告指示或其它指示。 A / R服務2 4也可提供消除岐義的組態問題指示。如 果幫助呼叫者1 2區別呼叫者1 2所蒐尋之資訊的消除、岐義 提示無法成功地提示呼叫者1 2,則產生警告。例如,如果 消除岐義的提示請求呼叫者12指出其所蒐尋之人的部門, 但呼叫者1 2無法決定其所要找的人是在那個部門工作,則 此失敗(如等待回應的時間結束)即被注意。此外,如果消 除岐義導致不正確的人被識別,指示也會被儲存及報告。重 複失敗的報告有助於偵測所提供之不適當的消除岐義資訊。 A / R服務2 4可經由安全的通信(諸如安全Η T T P 傳送或SMT Ρ郵件)接收代表呼叫者發言記錄的資料、事 件記錄、以及其它記錄 '統計等。發言的記錄及S L Ε Ε記 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公爱) (請先閱讀背面之注意事?ΡΪ:填寫本頁) 裝 訂: 經濟部智慧財產局員工消費合作社印製 -42- 474090 A7 _B7 五、發明說明(40) 錄可供A / R服務2 4用來識別、調整及監視。I v R 1 6被架構成周期性地送出代表發言的記錄及s L E E記錄 的資料給A / R服務2 4。 A / R服務2 4也監視包括在引擎系統3 0中之識別器 3〇2的性能。例如,A / R服務2 4可以使用已知的測試 序列執行離線的識別測試。 A/R服務2 4也被架構成更新管理系統3 2中的資訊 。A / R服務2 4可以增加或移除字的發音,以及增加字典 中的名稱或字。此外,A / R服務2 4可以修改:I V R 1 6中所使用的Backus-Naur Form(BNF)文法。此有助於處理 諸如"Mike Phillips,please.” 之類的發言。A/R 服務 2 4 也可增加或更新聲音模型、識別器參數及語意模型(例如先 前之名稱的機率)。執行時間系統的升級及更新可以由A / R服務2 4執行。此外,A / R服務2 4被架構成經由介面 4 0控制波形及組態記錄的數量。此項控制包括開或關、閉記 錄波形,以及從記錄取樣的波形切換到記錄所有波形,或只 記錄錯誤的波形。 * 經濟部智慧財產局員工消費合作社印製 A / R服務2 4被架構成根據監視I V R系統1 6所得. 到的各種警告及警報採行適當的支援動作。A/R服務2 4 被架構成能將壞的通信線路變成忙碌的固定狀態。A / R服 務2 4也能重新開始部分的I V R系統1 6,收集長檔案供 除錯,以及將補綴的組態檔插入損壞的組態檔中。 現請參閱圖4,A / R服務24被架構成服務多個分散 的I V R系統。如圖所示,A / R服務2 4可以經由網際網_ -43- ----5---.-------裝—— (請先閱讀背面之注意事0填寫本頁) 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) 474090 A7 B7 五、發明說明(41) (請先閱讀背面之注意事填寫本頁) 路2 2、S Μ T P伺服器1 8及防火牆2 0服務I V R系統 16,以及IVR系統120、122及124 ’系統 120、122、124是爲不同的各個公司所架構。 A / R服務2 4是經由網際網路2 2及防火牆1 2 6及 SMTP伺服器128服務IVR系統12 0、122及 1 2 4。因此,如圖所示,A / R服務2 4經由多個 SMTP伺服器18、128服務IVR系統16、120 、122及124,以及,經由一個SMTP伺服器128 服務多個I V R系統1 2 0、1 2 2及1 2 4 ° : I V R系統 16、120、122及124可能相互分散在各處。 諸如電子郵件或S Ν Μ P的警告,可經由A / R服務 24傳送給實體,如SWI ,或其它的IVR 16供應商 ,或其它實體,以分析決定可能的改正動作。A/R服務 經濟部智慧財產局員工消費合作社印製 2 4被架構成當性能偏離被.監視之統計的預期行爲(例如一 預期値,例如是頻率或量)一既定量時,送出警告。AR 服務2 4中包括一架構實體(configuring enUty),架構實體 中包括轉錄器(transcriber) ,用以根據A / R服務2 4 警 告及警報或其它通知轉錄所儲存之呼叫者的發言。位在架構 實體的人檢查這些被轉錄的發言,將這些發言與字典比較, 或分析轉錄的發言以決定適當的動作,如果有。此動作包括 使用R C S 3 6調適/重新架構I V R 1 6,例如,降 低〇0 V s ,更新發音或其它資訊,及/或修改儲存在 I V R 1 6中的資訊。 -44- 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) 474090 A7 —______ B7 五、發明說明(42) 架構與建立 · 如何架構與建立系統1 〇,視客戶所選擇的系統類型而 定。客戶可以選擇基本平台及架構工具,或一組可架構的模 型。如果客戶選擇基本平台及工具,則客戶可架構及自訂產 品。客戶可以提供架構/客戶資料給供應商及/或架構實體 ’例如Speech Works® International Inc.,供架構實體架構 系統1 0。 如果客戶選擇基本平台及工具,則客戶爲所要的功能及 任何客戶自訂參數輸入資料。客戶需要經由整個:資料庫的下 載或個人的登錄輸入自動陪從的相關資訊,如人員姓名、暱 稱、部門、及分機號碼,以及任何適當的同義字,如職銜/ 職稱。此外,客戶可以提供語音頁內容的資訊,以及鏈結到 其它頁的指令,以及語音頁所支援之交易的指令。基本平台 內提供了很多內容及功能,包括交易,但客戶需要提供客戶 專有的資料及指令。客戶可以選擇資訊組態參數以將系、統的 性能客戶化,例如,自動陪從時是否可以消除岐義。如另一 例,客戶可使用事件登記工具將日期、事件名稱、以及呼叫 者所需要的提示資訊記錄到暫存器中用於即將發生的事件。 客戶可以相同的方式按需要修改/更新初始的架構/建立, 以適應改變的資訊,如最近的事件,人員的增減,名稱的改 變,或事件中其它改變等。 如果客戶選擇可架構的模型組由供應商或其它實體架構 ,則客戶應提供相關的資訊給架構實體,例如Speech Works® International Inc·。客戶應提供語音頁的內容資訊,— 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) (請先閱讀背面之注意事?^:填寫本頁) 裝 n n I n n n n n 1 -is 經濟部智慧財產局員工消費合作社印製 -45- 474090 A7 B7 五、發明說明(43) ——-----------裝—— (請先閱讀背面之注意事填寫本頁) 用於上述自動陪從的相關人員目錄資訊,以及用於組態參數 所想要的選項。架構實體使用此資訊以及它的專業爲客戶架 構系統。此外,在初始架構/建立後,架構實體可按照客戶 的需要更新架構/建立。 無論是客戶或其它實體執行架構,組態檔被寫入管理系 統3 2 (圖1 )及/或被其修改,並被引擎系統3 〇 (圖1 )讀取供執行。 操作 經濟部智慧財產局員工消費合作社印製 操作時,I VR系統1 6與使用者/呼叫者1 2間的互 動,是按照使用者介面經由網路模式的語音識別法引導呼叫 者1 2 ’其中包括執行呼叫者1 2所指示的操作。按照網路 模式,典型上,首先呈現給呼叫者1 2的語音頁是首頁(除 非呼叫者1 2直接存取不同_的語音頁),首頁中提供呼叫者 1 2各種選項。呼叫者1 2可經由說出任何指定的字片語 在所呈現的選項中選擇,或是以自然說話的語言說出呼叫者 1 2所想要的資訊及/或服務。所使用的術語類似網站所使 用,以有助於呼叫者1 2經由說出適當的發言導覽各語音頁. 。對呼叫者1 2的每一個發言而言,可以將系統架構成執行 消除岐義及/或再嘗試識別。I V R系統1 6與呼叫者1 2 間對話的每一階段,呼叫者1 2都被告知那一頁被載入(例 如”載入與我們聯絡頁”),且在語音頁被載入後,將所呈 現之頁的名稱告知呼叫者1 2 (例如”與我們聯絡頁,呼叫 我們免費......” )。A / R服務2 4分析及監視 - 鶚.. 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) -46 - 474030 A7 B7______ 五、發明說明(44) (請先閱讀背面之注意事填寫本頁) 1 VR系統1 6與呼叫者1 2間有關對話的資訊,並提供適 當的報告、警報及/或警告供客戶及/或架構實體(例如 Speech Works® International Inc·)使用,以決定系統 1 〇 是否要被更新。 現請參閱圖1、2及5,互動式對話方法2 0 0從階段 2 0 2呼叫者1 2撥出與I V R系統1 6相關的電話號碼開 始。呼叫者1 2從呼叫者的電話2 6經由PSTN 1 4連 接到I V R系統1 6。I V R系統1 6與呼叫者1 2間經由 PSTN 1 4建立連接,呼叫者1 2與I V R :系統1 6間 是雙向通信。 在階段2 0 4,I V R系統1 6播放提示給使用者1 2 以指示使用者1 2到達SpeechSite™IVR系統1 6的首頁。例 如,播放給呼叫者1 2的提示例如是”首頁”或”你已進入 〔某公司〕之SpeechSiteTMp音識別系統的首頁”。另者, 如果使甩者1 2所撥的號碼與某特定語音頁有關而非首、頁, 經濟部智慧財產局員工消費合作社印製 則所撥頁的資訊會提示/播放給呼叫者1 2。提示給呼叫者 1 2的資訊包括各種選項,以便使用者1 2存取其它頁及/ 或得到首頁上的一般資訊。提示可以告知呼叫者1 2與 SpeechSite™IVR系統有關的語音模組。在本例中,提示包括 ”請說‘公司目錄’即可鏈結到個人目錄;請說‘與我們聯絡’ 即可找到與我們聯絡的方法;請說‘公司資訊’即可瞭解本公 司;執行〔交易X〕請說〔執行交易X〕” 。交易例如是購 買股票或其它商品。因此,在本例中’〔執行交易X〕及〔 交易X〕都可以”買股票”。因此’提示中指示呼叫者1 2 一 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) ^7- ' 474090 Α7 Β7 五、發明說明(45) 如何開始呼叫,安排線路分別完成公司目錄(自動陪從)、 資訊擷取(與我們聯絡及公司資訊)及交易處理。資訊也包 (請先閱讀背面之注音?事1U:填寫本頁) 括如何導覽各語音頁的指令,給呼叫者1 2的提示包括呼叫 者1 2說出與網站功能類似的相關名詞即可導覽A This paper size applies to China National Standard (CNS) A4 specifications (210 X 297 public love) -26- 474090 A7 B7 V. Description of invention (24) Does anyone or department in the business correspond to the name, department or title? . The engine system 3 0 can disambiguate the response, for example, prompting the caller 1 2, in addition to the employee's name, 'the department of the employee must be confirmed. The function of arranging call lines also includes the execution engine 80 connecting the caller 12 to the requested person / department. In particular, the execution engine 80 is constructed using blind fl a s h-h ◦ k transmission, and calls are transmitted according to the data stored in the configuration and recording system 34. The engine system 30 is also structured to perform other types of transfers, such as supervised transfers. For the information retrieval function, the engine system 30 can identify the specified voice page requested by the caller 12 or decide which page contains the requested information. The engine system 30 is constructed to recognize the voice from the caller 12 in order to decide what kind of information the caller 12 requests. In response to the voice of the identified caller 1 2, the engine system 30 is framed to access the designated / determined page and plays a prompt related to the requested information to the caller. Therefore, when the engine system 30 is on the appropriate page (the page that provides a link to the "Contact Us" page), for example, it can link to the user / caller 12 saying "Contact Us" and link to "Contact Us" page. . In addition, when the engine system 30 is on the appropriate page (providing a page linked to the "Contact Us" page), for example, it can respond. The user requests information from the "Contact Us" page and links to the "Contact Us" page. For example, caller 12 says "direction to Boston" and engine system 30 links caller 12 to the "Contact Us" page. The engine system 30 is also framed so that callers 12 can execute transactions via a specific IVR system 16. In order to execute the transaction, the engine system 30 is framed to identify a designated page or decide to include the information requested by the caller 12 or-This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 public love)- -& --- · ------- install --- (Please read the note on the back 0 to complete this page) · Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs -27- 474090 A7 __ B7 V. Description of the invention (25) The page of the service (such as stock trading) and accesses the designated / determined page, plays a prompt to the caller 12 and performs / initiates the function designated by the caller 12. The engine system 30 recognizes the caller's voice and associates the recognized voice with the data stored in the configuration and recording system 34. According to the instructions stored in the engine system 30 and / or the data (including appropriate instructions) stored in the configuration and recording system 34, the engine system 30 can execute the instructions according to the information provided by the caller 12 Transaction. To control the conversation with the caller 12, the engine system 30 interacts with the caller: 12 via the PSTN 14 and the configuration and recording system 34. The engine system 30 can receive the voice of the caller 12 via P S T N 1 4. To recognize speech, the engine system 30 can access information in the configuration and recording system 34 indicating which speech can be recognized by the IVR system 16. The engine system 30 is structured under the control of the execution engine 80 to manage the conversation between the I VR system 16 and the callers 1_2. Depending on whether the engine system 30 recognizes the voice of the caller 12, the execution engine can instruct the engine 30 to output the prompts stored in the configuration and recording system 34 to the caller 12. These prompts can be, for example, according to the previously recognized voice, # 叫 者 1 2 requested information, ask the caller 12 to speak again an unrecognized voice or a recognized but low-reliability voice, or to the engine system 3 0 Non-voice information received sends other appropriate error messages. The engine system 30 is structured to communicate with the caller 12 in a conversation-oriented manner to guide the caller 12 to obtain the desired result of the caller with a limited vocabulary. The engine system 30 presents commands that callers can say, and uses recognized vocabulary, including possible commands (such as "contact us"). This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 public) (Please read the notes on the back to fill in this page first)-Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs -28- 474030 A7 B7 V. Description of the invention (26), and some synonyms or other transmissions have the same meaning (Such as "direction"). Approved vocabulary changes depending on the state of the interactive conversation with the caller. Caller 12 can also say "full" or "universal" commands for any voice page, such as "description", "previous page", "next page". The "previous" and "next" commands only apply if the current voice page is still before or after the voice page that the caller 12 has entered. Speaking outside the allowed 'callers 1 2 will get error messages. Limiting the available recognizable speech helps to increase the accuracy and speed of recognition and the overall stability of the I VR system. : Engine system 30 is framed to identify and respond to caller 12 requests using conventional techniques. For example, speech from caller 12 can be parsed into speech units and converted by a digital signal processor to produce a vector of speech units. These vectors are grouped into speech segments of different lengths. These segments are then converted into feature vectors, and these feature vectors are then analyzed by language restrictions (such as recognized vocabulary) to produce an N-best table of N strings with the highest trust score. The interaction between the engine system 30 and the caller 12 is presented through a user interface, which converts the visual format provided on a typical website into a sound format, including the functions provided by the website browser. The user interface is completed by the prompts accessed by the engine system 30 and presented to the caller 12. The prompts played by the engine system 30 are stored in the configuration and recording system 34. The same information is provided to the person browsing the website, for example, to inform the caller of the current location (such as "Home") or the location to be moved (such as "Move to Contact Us Page"). These tips for users use website-related terms, such as "page", "contact us", "company information", "back to the page ... (Please read the notes on the back first and fill in this page ) I printed this paper printed by the Consumers' Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs. The paper size is applicable to the Chinese National Standard (CNS) A4 (210 X 297 mm) -29- 474090 A7 B7 V. Description of Invention (27) Go to page ·, ". In other examples, the prompt played to caller 12 may be "You can say [Page 1] [Page 2] ... (Please read the note on the back HI and fill out this page)". [Page 1] [Page 2] The prompts are structured text prompts, such as customer text stored in the configuration and recording system 34, and retrieved and played back to the caller 12. These words replace the words in the greeting. For example, the engine system 30 provides the client text to the caller 12 immediately based on the information stored in the configuration and recording system 34. Typical customer pages are, for example, company-specific products, company-specific departments, and / or company-specific services. Caller 12 can also be presented with prompts for other queries, such as asking yes / no. The organization of the information presented to the caller 12 by the engine system 30 may be different from the content of the corresponding website, but basically the same information is presented to the caller 12. As part of the user interface of the homogeneous-website, the engine system 30 is structured to respond to the order of the web-site to navigate the entire information in the IVR system 16. The caller can say the typical commands provided by the web browser, such as "first page", "previous page", "next page", "description" and "to", and the engine system 30 is constituted to recognize these commands And take action. Therefore, the engine system 30 responds to the above-mentioned command of the caller 12 and moves the caller 1 2 from the page to the previous page, the next page, or returns to the home page of the Ministry of Economy intellectual property in the page: Printed by the bureau's consumer cooperative. For each voice page, they can be linked to a specific page that the engine system 30 prompts the caller 12. For example, the homepage may have special links to company information pages, contact us pages, and product / service pages and present the links to callers 1 2. For example, callers 12 can be told "you can say" company information "," contact us ", or" products and services "". Engine Department _ -30- This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) 474090 A7 --- B7 V. Description of the invention (28) System 3 0 can be moved as requested by caller 1 2 To these specific pages. 、 ------- * ------- install --- (Please read the notes on the back β to complete this page) The engine system 3 0 is also structured to provide search services and can automatically fax the service To caller 1 2. At the appropriate point in the conversation with the I v R system 16, caller 12 can say "look for". In response to this request / speech, the engine system 30 searches the stored voice page for the indicated text and / or information. The caller 12 can say "Fax it" at any time, and the engine system 30 is structured to respond to this request and fax the contents of the current page (inner valley related to the current voice page) to the caller 12 designated Fax number. The fax number can be stored and confirmed first, or the engine system 30 can request the fax number after responding to the fax command. Printed by the Consumer Property Cooperative of the Intellectual Property Bureau of the Ministry of Economics The engine system 30 is constructed to record call events and other transactions into the configuration and record system 34. Call events are the phases of the interactive conversation between the IVR system 16 and the caller 12. These events include the request of caller 12, the recognition made by engine system 30, indicating whether the caller's voice is recognized, whether the voice is not accepted because of low recognition reliability or not recognized, Start the action, and play the prompt to the caller 1 2. The event may include the page that the caller wants, the order / order in which the caller 12 requested, and the action performed by the engine system 30. The engine system 30 is configured to transmit the call event indication to the configuration and recording system 3 4 for storage for future reference. The engine system 30 is framed to send out instructions for certain call events whenever a call event occurs. At the same time, instructions for other call events are transmitted only when certain conditions occur, and instructions can be framed to not send out other call events at all. For example, the engine system 30 can send an instruction every time a low level of reliability is not accepted, or a high level of reliability is accepted (for example, a certain -31-this paper size applies to the Chinese National Standard (CNS) A4 specification (210 X 297 mm) (%) 474030 A7 ___ B7 V. Description of the invention (29) (Please read the notes on the back and fill in this page first) People successfully connected using the auto-accompaniment feature>. Engine system 3 〇 It is also constructed to make calls based on call events Reports are generated for statistics and stored in the configuration and recording system 34, and / or retrieved by the monitoring interface system 40. Configuration and recording system configuration and recording system 34 includes a record storage area 8 6. Database storage Area 8 8 and general storage area 90. The configuration and recording system 3 4 is framed to form the storage management system 3 2, the engine system 30, the support system 3 8 and the monitoring interface system 4 0, and the information used with these Each of the systems is a two-way interaction. Therefore, each of these systems 30, 32, 38, 40 can retrieve information from the configuration and recording system 34, or store information in it. 0 Intellectual Property Office staff, Ministry of Economic Affairs Consumer Cooperation The company prints a database 8 8 to store the configuration file and the content of the voice page. The content of the voice page is similar to the content and format generally obtained from the website. The configuration file is used to structure the system 30, 32, 36, 38, 40 During the construction and establishment period, the information required to construct each of these systems 30, 32, 36, 38, 40 is stored in these configuration files. These files can be made by the manufacturer and / or purchaser of the IVR system. Create and / or modify to provide and / or change the customer's configuration. The database 88 is also structured to store various information related to voice pages. For example, the database 88 is structured to store information related to prompts. The reminder information includes identification, recording date, name of the person who recorded the reminder, source, and type of reminder. In addition, whether the reminder is made public, the unique user interface name used for the reminder, and the text of the reminder are stored in the database 88. In addition, the location of the reminder in the database 8 and the date when the reminder was generated are also stored. -32- This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) 474030 A7 B7 Ministry of Economy Wisdom Printed by the Production Bureau Staff Consumer Cooperatives V. Invention Description (30) is stored in the database 88. Information that links multiple reminders to form phrases or other reminder segments is also stored in the database 88. 1 The database 8 8 also stores information related to the voice module. This information includes ... . . . . . . . . . . _________________________________________ Identification information for voice modules and voice pages, including the content of each module and page. This identification information is suitable for the engine system 30 to locate, retrieve and process various voice modules and voice pages, and prompts are also included in it. The database 88 also stores data related to the voice page. Page-to-page links are included in the database and help the engine system 30 link to other pages and / or modules to more easily retrieve information and perform other actions. The database 88 also stores the information of the DialogModule ™ voice processing unit 300 (Figure 10) and the link to a specific voice page and / or voice module. 〇 ^ 1〇 £ 乂 (^ 111 ^ Qiaowu sound processing unit 3 0 0 (Figure 10) The link information provides a mapping to determine which DialogModule ™ speech processing unit 3 0 0 is to be executed when the voice page is executed. (Figure 10). The data stored in the database 88 is also provided as synonymous information and links between materials. When synonymous words are available for an item, it will help increase the accuracy of identifying the item ”Database 8 8 also stores several other types of information. This information includes information that helps support audio page navigation, including navigation nouns linked to navigation functions, and when implemented in the engine system 30 The keyword used to locate the voice page when the engine 80 performs the "find" function. The user dictionary is also stored in the database 88. The database 88 also contains information related to the operation of the company. For example, the date the company opened and / Or the time is also stored in the database 88. The database 88 also stores the information of the personal directory to provide the auto-accompaniment function. Stored in (please read the notes on the back first and fill in this page) Paper size applies Chinese National Standard (CNS) A 4 Specifications (210 X 297 mm) -33- 474090 A7 B7 V. Description of the invention (31) (Please read the notes on the back first 填写: Fill out this page) The information in the personal directory in the database 8 8 is stored in the data Columns. Blocks include, for example, the names of people, nicknames, titles, departments, synonyms for the registration of these blocks, and the extensions of individuals, offices, departments, or other lines that transfer / arrange call lines to individuals and / or departments. Information. These fields can be updated to reflect new entrants, changes in departments, changes in department names, changes in names and personnel, and addition of nicknames or other synonyms. 'R The content of the stored voice page contains each of the IVR system 16 All information on one page, including tips (such as inquiries and information), configuration and 1 knots > words # The content of the page is divided into columns that can be selected in order to 'page the page before sending it to the buyer / customer The content of the page is modified to the structure required by the customer, or it is modified by the customer of the IVR system 16. The content of the voice page can be updated as needed by modifying the data columns, for example, updating the stock price, providing Daily news, or point out any changes that may occur to the company. The Intellectual Property Bureau, Ministry of Economic Affairs, Employee Consumer Cooperatives printed storage area 90 stores all reminders, fax pages, GIFs, and voice models. All reminders are sound information for use Rebroadcast to caller 2. For example, the prompt includes a call to caller 12 and a statement of information sent to caller 12. The fax page is a response to a request for caller fax information (for example, call til [caller 12 said "fax it", etc.) ) Data transmitted to caller 12's facsimile machine 2 8 (Figure 1). Graphic information is a structured file of Graphic Interchange Format (GIF) and can be included in the fax page. The speech model is used by the engine system 30 to recognize part of the speech to recognize the words and / or phrases spoken by the caller 12. The record storage area 86 is constructed to store records of call events and other information required by the system 40, such as service logic and execution environment (SLEEE) records. The record of the call event includes relevant statistics. For example, the paper size of the call is applicable to the Chinese National Standard (CNS) A4 specification (210 X 297 mm) 474090 A7 B7 V. Description of the invention (32) The call length r request Voice page I, recognition success rate, words other than recognition, failed recognition, and commands used. Support system The support system 38 is constructed to be requested by the management system 32 and / or the engine system 30 and provides the support required by these systems 30 and 32. The support system 38 includes a text-to-speech (TTS) feature 9 2. a record converter 94 4 'fax feature 96, a report generator 98 and a voice adaptor 100. T T S 9 2 allows the engine system 30 to output speech or other appropriate sounds to the caller 12 based on the text stored in the I VR system 16. The implementation of T T S 92 can use well-known techniques, such as the Lucent T T S bow. T T S 9 2 allows the I VR system 16 to be updated very quickly. For example, press releases can be quickly stored in text to the configuration and recording system 34, using T T S 9 2 and the engine system 3 0 can be immediately spoken. The tone is output to the caller 1 2. For example, it can be recorded into a recording by a famous news anchor later, which can be used instead of T T S 9 2 to convert text news into speech suitable for engine system 30. Other parts of I V R system 16 can also be updated in this way, for example. For example, the employee roster in the personnel directory. The record converter 94 is constructed to convert the information of the records stored in the storage area 86 into an appropriate format for processing by the report generator 98. Here, the record converter 94 is configured to access the SLEE files stored in the storage area 86 and convert these files into NC SA (National Center for Supercomputing Applications) standard records. Record converter 9 4 OK-This paper size is in accordance with Chinese National Standard (CNS) A4 (210 X 297 mm) (Please read the notes on the back first and fill in this page) -35- 474090 A7 _ B7 V. Description of the invention (33) Effectively convert the caller's access instruction to the IVR system 16 into a "hit" equivalent to the web page, and store these finds in the file . Therefore, the record converter 94 is structured to store a file containing the ID (such as a phone number) of the caller 12, the date and time of the caller 12 request, and the information of the caller 12 request for information or action. Instructions. The records stored by the record converter 94 are stored in the configuration and recording system 34. The facsimile feature 96 is constructed to process a facsimile request from the caller 12 and fax the requested information to a facsimile machine 2 8 available to the caller 12 (Figure 1). For example, fax feature 9 6 can use WmFax Pro 9. 0 implementation. This implementation of the facsimile feature 96 supports RightFax servers and Internet servers. The facsimile feature 9 6 can fax information to a facsimile number associated with the facsimile machine 2 8. The facsimile number is provided by the caller 12 via the facsimile server 9 7. The report generator 98 is configured to access records and other information stored in the configuration and recording system 34, and operate these data to generate various reports and reports. For example, the report generator 98 may operate the record stored by the record converter 94 to generate a report related to the voice page that was found. The report generator has been constructed from 8 frames to generate reports indicating the number of calls per hour, the report of the number of calls per hour for all voice modules, and the report of the number of operator transmissions per hour. The report generator 98 can also generate a report of the number of calls from a specific device identified by the automatic number identifier (A N I) on the selected day / week / month. These reports can be written and drawn in a format that can be downloaded and imported into the database. The voice adaptor 1 0 0 is framed for the adaptation of the engine system 3, 0 _ -------- * ------- install --- (Please read the precautions on the back first '' (This page) Printed by the Employees 'Cooperatives of the Intellectual Property Bureau of the Ministry of IT and the Ministry of Economics This paper is printed in accordance with the Chinese National Standard (CNS) A4 (210 X 297 mm) -36- 474090 A7 B7 Printed by the Employees' Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs V. Description of the Invention (34) Tool 'to help improve the speech recognition of the engine system 30. Speech Adaptor 1 0 0 Speech Works® Inter * riati〇nal Inc. LEARN 6 · 0 software implementation. The speech adaptor 100 can access the information stored in the configuration and recording system 34, analyze this information, and decide how to adapt the sound model, pronunciation curve diagram, and / or stored in the configuration and recording system 34. Semantic model. The speech adaptor 100 is also constructed to update / change the sound model, pronunciation graph, and / or semantic model according to the decision made. The new model and graphics are then stored in the configuration and recording system 34, which is used by the engine system 30 to recognize the voice from the caller 12. : Remote control system The remote control system (RCS) 3 6 is constructed to provide IVR 16 remote control via the analog communication line 104. RCS 36 includes Remote Access System (RAS) 106, which is controlled by appropriate software, and is PC Anywhere 108 in this document. RAS 1 0 6 communicates with pro-line 1 104 via modem 1 1 0. R C S 3 6 allows control over I V R 1 6 through the NT window. For example, RCS 36 can be configured to allow start / stop processing to modify the configuration of system 30, 32, 38, 40, including the data stored in it, access management system 32, and start / Close the communication line connected to IVR 16. Surveillance interface system Surveillance interface system 40 provides monitoring functions for I V R 16, including. ----. ---. ------- Packing --- (Please read the notes on the back first_fill in this page) · Thread-This paper size is applicable to China National Standard (CNS) A4 (210 X 297 mm) -37- Economy Printed by the Consumer Cooperatives of the Ministry of Intellectual Property Bureau 474030 A7 ________ B7 V. Description of the Invention (35) Including system monitor 1 1 2, prompt monitor 1 1 4 and adjustment monitor 1 16. Each of these monitors 1 1 2, 1 14 and 1 16 is framed in the form of a u 1 aw file (μ— 1 aw file 'text waveform) to retrieve information from the configuration and recording system 3 4 Store in it 'and communicate with the SMTP server 18 bidirectionally. The reminder monitor 1 1 4 constitutes a monitoring reminder for changes and provides an alert when the change occurs. The system monitor 1 1 2 is framed to monitor the computer functions 1 v R 1 6 and responds to the monitoring function to take appropriate action and provide a "base heartbeat" to the A / R service 2 4 ( figure 1 ). : Basic heartbeat is a message to A / R service 2 4 telling A / R service 2 4 I V R 1 6 to operate and operate with normal operating parameters. The system monitor 1 1 2 provides alerts and warnings to the hardware, as well as phone errors, resource constraints, execution time errors, and / or transaction errors. A resource limitation error is applied to the code of the application software in I V R 16. Execution time errors are provided to SLEE, speech recognizers, and DialogModule ™ speech processing orders and meta libraries. The SLEE library is constructed to accept calls from the callers 12 and request the engine system 30, including its speech recognizer. The execution time and transaction errors in the IVR ‘16 software code include all kinds of errors encountered when processing calls from callers 1 2. System Monitor 1 1 2 also reports transaction errors and indicates these transaction errors. Stored in the configuration and recording system 34. System monitor 1 1 2 is also configured to perform certain remedial actions, such as restarting non-critical services that select I V R 1 6. Alarms and warnings sent by the system monitor 1 1 2 are sent to the a / R service 2 4 (Figure 1) via the Internet (Figure 1). This paper size is in accordance with Chinese National Standard (CNS) A4 (210 X 297 mm) -38- (Please read the notes on the back first? ^ Fill in this page) »1 474090 A7 B7 V. Description of the invention (36) (Please First read the notes on the back and fill in this page) ^ Monitor 1 1 6 § The frame structure monitors and analyzes the voice performance of the interaction between caller 12 and IVR 16. The adjustment monitor 1 1 6 is constructed to calculate performance statistics from S L E E records stored in the configuration and recording system 3 4 and track performance statistics. From the performance statistics, adjusting the monitor 1 1 6 can send a warning about these performance statistics. The adjustment monitor 1 1 6 also sends the S L E ′ E record and the waveform from the caller 12 to the portion indicated as a possible problem in the voice for external monitoring. Adjusting the monitor 1 1 6 can also output status messages about chat statistics. Adjust the monitor 1 1 6 to transmit these warnings, records, waveforms, and messages to the A / R service 2 4 (Figure 1) over the Internet 2 2 (Figure 1). Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs. The adjustment monitor 1 1 6 is constructed to provide many reports related to performance statistics of conversations between callers 12 and I R 16. Adjusting the monitor 1 1 6 can analyze performance statistics according to several criteria, such as the transaction completion rate of important transactions, the DialogModule ™ voice processing unit completion rate, failed calls, the response time felt by callers, the proportion of short calls, and the cut of calls Breaks, the number of relative transfers, the number of calls transferred to the agent, and the volume of the calls. That transaction should indicate that the important transaction was after the system was erected or modified. DialogModule ™ voice processing unit completion rate information includes how many confirmations occur and how many failures occur. Information on the completion rate of the DialogModule ™ speech processing unit is formatted according to the speech page associated with the DialogModule ™ speech processing unit 3 0 0 (Figure 10). The response time felt by the caller can be used to determine if the work of the I R 16 is overloaded. The time the person's name has not been used is determined in advance and can be selected as needed, such as 1, 6 and / or 12 weeks. About the number of cut-off and relative transmission of calls and transmission to customer service _ -39- This paper size applies to China National Standard (CNS) A4 (210 X 297 public love) 474030 A7 B7 V. Description of invention (37) Number of calls by members The report is useful for analyzing auto-accompaniment performance. (Please read the notes on the back first and fill in this page) Adjusting the monitor 1 1 6 also provides several business reports. For example, it provides reports on the number of calls per hour, the number of calls per hour on important conversations, the number of calls per hour from a customer service agent, and the number of calls from an AN within a predetermined period. The number of calls per hour is provided in a downloadable format, such as text and graphics. Important dialogue paragraphs are defined by configuration files stored in the configuration and recording system 34. A N I reports a given time frame such as day, week, or month. These reports are provided in an exportable format, via a text file or loaded into a database for business information and other reporting functions. Adjust monitor 1 1 6 to trigger alarms under various failure and / or performance conditions. These alerts are sent as structured messages to one or more destinations using, for example, a single network management protocol (SNMP) or email. Alerts help off-sheer monitoring and / or architecture I V R 16 of system performance through customer operations centers, off-site external resource monitoring companies, and / or sales entities. Analysis / Reporting Services. Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs. The adjustment monitor 1 1 6 can send the report to the A / R service 2 4 (Figure 1) via the Internet 2 2 (Figure 1). Referring again to Figure 1, the A / R service 2 4 is constructed to monitor the performance of I V R 1 6 and provide warnings to initiate diagnostic actions regarding I V R performance. Diagnostic actions can be performed, for example, by a supplier of I v r 1 6 such as Speech Works® International Inc. . A / R service 2 4 frame constitutes access to the data in the configuration and recording system 3 4. Analyze these data and determine appropriate actions to generate appropriate warnings. -40- This paper standard applies Chinese National Standards (CNS) A4 specification (210 X 297 mm) 474090 A7 — B7 V. Description of the invention (38) and / or generate appropriate report ~ A / R service 2 4 The quilt constitutes periodic access and / or reception of data, such as storage Records of speeches and SLEE records in the configuration and recording system 34 are used to identify, adjust, monitor and generate reports. One type of alert that A / R Service 2 4 can generate is a potentially high 00V rate. Reasons that can cause a high 00V rate are, for example, the name list stored in the personnel directory of the configuration and recording system 34 is not maintained. Therefore, when caller 12 requests a specific name, 1 VR 16 does not accept the requested name, even though the requested person is an employee of the company served by I V R 16. A / R service 2 4 may generate warnings and / or reports for possible candidate synonym / nickname identification. Unrecognized phrases or phrases with low reliability are accepted for recognition (with high reliability) when they try again (for example, caller: "OEC of your company"; IVR: "I do n’t understand, please Say name "; Caller:" Stuart Patterson / '), the phrase used by caller 12 in the first attempt is a good candidate, synonym for the person identified in the second attempt. A / R Service 2 4 can generate a report indicating possible synonyms (eg CE0) and recognized speech (eg Stuart Patterson). A / R service 2 4 can also generate warnings for repeated bad pronunciations. What callers 12 say A high percentage of recognition for a phrase indicates a poor pronunciation of the phrase related to the planning of I VR 16. The warning indicating repeated recognition of the recognition word / phrase can be used to initiate an action to adjust the pronunciation of IVR 1 6 Recognition to reduce the number of approvals required for a word / phrase. A / R Service 2 4 is also structured to generate alerts, warnings and / or reports for names that are not recognized by I VR 1 6. Names with a high previous trust score 1 This paper size applies to Chinese national standards CNS) A4 specification (210 X 297 gong) (Please read the note on the back and fill in this page first) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs -41-474030 Α7 _ Β7 V. Description of Invention (39) The low trust score indicates several reasons, such as poor pronunciation recognition and / or high level of noise and / or the person is a former employee of the company. / R service 2 4 frame constitutes the recognition threshold for monitoring the trust distribution to help manage trust scores. IVR 1 6 can also be adapted to enhance caller's speech recognition to increase the accuracy of recognition. By doing so, you can shift all The entire distribution of confidence scores for speeches. However, a decrease in the percentage of speeches that are not accepted may indicate an increase in false acceptance (ie, unrecognized ones are also accepted) because the threshold for acceptance is too low. Conversely, if the threshold for rejection is not Too: High, the rejection rate is artificially high, which suppresses the true recognition accuracy. A / R service 2 4 can monitor the distribution of trust and influence the threshold to help obtain the appropriate threshold to help achieve true of Different accuracy. A / R can also generate warning indications or other indications of personal trust scores and non-acceptance rates. A / R service 2 4 can also provide disambiguation of configuration problem indications. If you help callers 1 2 differentiate calls Elimination and disambiguation of information searched by person 12 fails to prompt caller 12 successfully, for example, if the disambiguation prompt requests caller 12 to indicate the department of the person they are searching for, but the caller 1 2 It is impossible to decide which person he is looking for is working in that department, and the failure (such as the end of the waiting time for the response) is noted. In addition, if the disambiguation results in the identification of an incorrect person, the instructions are also stored and reported. Repeating failed reports can help detect inappropriate disambiguation information provided. The A / R service 2 4 can receive information on the caller's speaking record, event record, and other records' statistics, etc. via secure communications (such as secure Η T T P transmission or SMT P mail). Speech transcript and SL Ε Ε notebook paper size applicable to Chinese National Standard (CNS) A4 specification (210 X 297 public love) (Please read the note on the back first? Filled in this page) Binding: Employees of Intellectual Property Bureau Printed by Consumer Cooperatives-42- 474090 A7 _B7 V. Description of Invention (40) The record can be used by A / R service 2 4 for identification, adjustment and monitoring. I v R 1 6 was constituted to periodically send the record of the representative's speech and the data of s L E E to the A / R service 2 4. The A / R service 24 also monitors the performance of the identifier 302 included in the engine system 30. For example, A / R service 2 4 can perform offline recognition tests using known test sequences. The A / R service 2 4 is also structured to constitute the information in the update management system 3 2. A / R Service 2 4 can add or remove the pronunciation of words, as well as add names or words in the dictionary. In addition, A / R service 2 4 can be modified: Backus-Naur Form (BNF) grammar used in I V R 16. This helps with things like " Mike Phillips, please. ". A / R service 2 4 can also add or update the sound model, recognizer parameters and semantic model (such as the probability of the previous name). The upgrade and update of the execution time system can be performed by the A / R service 2 4 Execution. In addition, A / R service 2 4 frame is configured to control the number of waveforms and configuration records via interface 40. This control includes turning on or off, closing the recorded waveform, and switching from recording the sampled waveform to recording all waveforms. Or just record the wrong waveform. * The A / R service printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 2 4 quilts constituted according to the surveillance IVR system 1 6 All warnings and alarms that have been detected will be appropriately supported. The A / R service 2 4 is constructed to be able to turn a bad communication line into a busy fixed state. A / R service 2 4 can also restart part of the I V R system 16, collect long files for debugging, and insert patched configuration files into damaged configuration files. Referring now to Figure 4, the A / R service 24 is framed to serve multiple decentralized IVR systems. As shown, A / R services 2 4 can be accessed via the Internet _ -43- ---- 5 ---. ------- Installation—— (Please read the note on the back of the page 0 to complete this page) This paper size is applicable to China National Standard (CNS) A4 (210 X 297 mm) 474090 A7 B7 V. Description of the invention ( 41) (Please read the notes on the back to fill in this page first) Road 2 2, SM TP server 18 and firewall 2 0 service IVR system 16, and IVR system 120, 122 and 124 'System 120, 122, 124 is Structured for different companies. A / R service 24 is serving IVR system 12 0, 122 and 1 2 4 through Internet 2 2 and firewall 1 2 6 and SMTP server 128. Therefore, as shown in the figure, the A / R service 24 serves the IVR systems 16, 120, 122, and 124 through a plurality of SMTP servers 18, 128, and serves the plurality of IVR systems 1 2 0, 1 through an SMTP server 128. 1 2 2 and 1 2 4 °: IVR systems 16, 120, 122, and 124 may be scattered among each other. Alerts such as email or SMP can be transmitted to entities such as SWI, or other IVR 16 providers, or other entities via A / R service 24 to analyze and determine possible corrective actions. A / R service Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 2 4 When the expected behavior of the monitored statistics (for example, an expected 値, such as frequency or quantity) is a given quantity, a warning is issued. The AR service 24 includes a structured entity (configuring enUty), and the structured entity includes a translator (transcriber) to transcribe the stored caller's speech based on the A / R service 2 4 warnings and alarms or other notifications. The person at the architectural entity examines these transcribed speeches, compares them to a dictionary, or analyzes the transcribed speeches to determine the appropriate action, if any. This action includes adapting / restructuring I V R 16 using R C S 36, for example, reducing 00 Vs, updating pronunciation or other information, and / or modifying information stored in I V R 16. -44- This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) 474090 A7 —______ B7 V. Description of the invention (42) Structure and establishment · How to structure and establish the system 1 〇, depending on the customer's choice Depending on the type of system. Customers can choose from a basic platform and architecture tools, or a set of architectable models. If the customer chooses the basic platform and tools, the customer can structure and customize the product. Customers can provide architecture / customer information to suppliers and / or architecture entities ’such as Speech Works® International Inc. For the architecture entity architecture system 10. If the customer chooses the basic platform and tools, the customer enters data for the desired function and any customer-defined parameters. The customer needs to enter auto-accompaniment related information such as the person's name, nickname, department, and extension number through the download of the entire: database or personal login, as well as any appropriate synonyms, such as title / title. In addition, customers can provide information on the content of the voice page, as well as instructions to link to other pages, and instructions for transactions supported by the voice page. A lot of content and functions are provided in the basic platform, including transactions, but customers need to provide customer-specific information and instructions. Customers can choose information configuration parameters to customize system and system performance, such as whether the ambiguity can be eliminated during auto-accompaniment. As another example, the customer can use the event registration tool to record the date, event name, and reminder information required by the caller into a register for an upcoming event. Customers can modify / update the initial structure / establishment in the same way as needed to accommodate changed information, such as recent events, personnel additions and deletions, name changes, or other changes in events. If the customer chooses an architecture group that can be architected by a vendor or other entity, the customer should provide relevant information to the architecture entity, such as Speech Works® International Inc. The customer should provide the content information of the voice page. — This paper size is applicable to the Chinese National Standard (CNS) A4 specification (210 X 297 mm) (Please read the notes on the back first? ^: Fill in this page) Install nn I nnnnn 1- is Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs-45- 474090 A7 B7 V. Description of the Invention (43) ——----------- Install—— (Please read the notes on the back and fill in this first Page) The relevant personnel directory information for the above-mentioned auto-accompaniment and the desired options for configuring the parameters. The architecture entity uses this information and its expertise to architect the system for the client. In addition, after the initial architecture / establishment, the architecture entity can update the architecture / establishment as required by the customer. Regardless of the execution architecture of the client or other entity, the configuration file is written into the management system 32 (Figure 1) and / or modified by it, and is read by the engine system 30 (Figure 1) for execution. When the consumer cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs prints and operates, the interaction between the I VR system 16 and the user / caller 12 is to guide the caller through the voice recognition method of the network mode according to the user interface. This includes performing the action indicated by caller 12. According to the network mode, typically, the voice page first presented to the caller 12 is the home page (unless the caller 12 directly accesses different voice pages). The home page provides various options for the caller 12. Caller 12 can choose any of the options presented by speaking any of the specified phrases, or speak the information and / or service desired by caller 1 2 in a natural speaking language. The terminology used is similar to that used by websites to help callers 1 2 navigate through voice pages by speaking the appropriate words. . For each speech of caller 12, the system framework can be configured to perform ambiguity and / or retry identification. At each stage of the conversation between the IVR system 16 and the caller 12, the caller 12 is informed which page is loaded (for example, "Load Contact Us Page"), and after the voice page is loaded, Tell the caller 1 2 the name of the page presented (eg "contact us page, call us for free. . . . . . ”). A / R Service 2 4 Analysis and Monitoring-鹗. . This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) -46-474030 A7 B7______ V. Description of the invention (44) (Please read the notes on the back first and fill in this page) 1 VR system 1 6 and Caller 12 information about the conversation and provide appropriate reports, alerts, and / or warnings for use by customers and / or infrastructure entities (such as Speech Works® International Inc.) to determine whether system 10 should be updated. Referring now to Figures 1, 2 and 5, the interactive dialogue method 2 0 0 starts from stage 2 2 2 caller 1 2 dials out the telephone number associated with the I V R system 16. The caller 12 is connected from the caller's phone 2 6 to the IV system 16 via PSTN 1 4. I V R system 16 and caller 12 establish a connection via PSTN 1 4 and caller 12 and I V R: system 16 communicate bidirectionally. At stage 204, the IVR system 16 plays a prompt to the user 12 to instruct the user 12 to reach the homepage of the SpeechSite ™ IVR system 16. For example, the prompt played to the caller 12 is, for example, "Homepage" or "You have entered the homepage of [Some Company] SpeechSiteTMp audio recognition system". In addition, if the number dialed by caller 1 2 is related to a specific voice page instead of the first page and page 1, printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs, the information of the dialed page will be prompted / played to caller 1 2 . The information reminded to caller 12 includes various options for user 12 to access other pages and / or get general information on the home page. Prompts inform callers about the 2 speech modules associated with the SpeechSite ™ IVR system. In this example, the prompts include "Please say 'Company Directory' to link to your personal directory; please say 'Contact Us' to find out how to contact us; please say 'Company Information' to understand our company; Execute [Transaction X] Please say [Execute Transaction X] ". Transactions are, for example, the purchase of stocks or other commodities. Therefore, in this example, '[Executive Transaction X] and [Transaction X] can both “buy stock”. Therefore, the caller is instructed in the reminder 1 2 A paper size is applicable to the Chinese National Standard (CNS) A4 specification (210 X 297 mm) ^ 7- '474090 Α7 Β7 V. Description of the invention (45) How to start the call and arrange the lines separately Complete company directory (auto-accompaniment), information retrieval (contact us and company information) and transaction processing. The information is also included (please read the phonetic on the back? Matt 1U: Fill out this page) Includes instructions on how to navigate the voice pages, and tips for callers 1 2 include callers 1 2 speaking related terms similar to the function of the website. Navigable
SpeechSite™ IVR系統如”上一頁”、”下一頁”及”首頁 ”,以及其它功能,如”尋找”、”傳真”及”我在何處? ” 〇 在階段2 0 6,呼叫者對電話2 6說出語音2 0 8以導 覽整個語音頁。語音2 0 8可以是回應I VR系:統1 6所播 放的提示,如請求指定的語音頁,也可以使用一般說話的語 言請求資訊或其它動作。語音2 0 8代表與語音相關的資訊 ,但不一定需要是類比或數位式的語音發言。例如,語音 2 0 8可以是識別器3 0 2輸出的一組N -最佳字串。 經濟部智慧財產局員工消費合作社印製 在階段2 1 0,引擎果統3 0辨別可用的子方法以決定 那一個子方法適合處理語音2 0 8所指示的請求。爲辨、別子 方法(每一個都包含一或多個語音模組),引擎系統3 0比 較語音2 0 8與提示給呼叫者1 2的子方法標題,及/或多 個字彙,這些字彙每一個都與可用的子方法至少其中一個有 關。在稍後的例中,字彙包含呈現給呼叫者1 2之標題的同 義字。例如,如果呼叫者1 2說出”到波士頓的方向”,方 法2 0 0將前進到階段2 1 4,以便從與我們聯絡頁中擷取 呼叫的資訊。如果語音2 0 8與子方法標題匹配(例如,如 果語音2 0 8是”公司資訊”),則引擎系統3 0指揮適當 的對應子方法處理語音2 0 8,例如在階段2 1 2的安排呼 本i張尺度適用中國國家標準(CNS)A4規格(21G X 297公爱) --- 474090 A7 _ B7 五、發明說明(46) 叫線路,在階段2 1 4的擷取資訊,及/或在階段2 1 6的 處理交易。 (請先閱讀背面之注音?事填寫本頁) 以下更詳細說明各子方法2 1 2、2 14、2 16對語 音2 0 8所做的處理。播放適當的提示給呼叫者1 2,以供 呼叫者1 2指示要指揮那一個子方法。例如,如果呼叫者 1 2要到安排呼叫線路子方法2 1 2,引擎系統3 〇會提示 ”轉移到公司目錄頁”(或”轉移到人員目錄頁”或”轉移 到安排呼叫線路頁”)。如果呼叫者1 2要轉移到子方法 2 1 4或2 1 6,則播放給呼叫者1 2的提示分1別是”轉移 到〔資訊擷取〕頁”及”轉移到〔交易處理〕頁”。在本例 中,以”與我們聯絡”或”公司資訊”取代〔資訊擷取〕, ”買股票”取代〔交易處理〕。另者,播放給呼叫者1 2的 提示還包括適當的頁已載入的指示,例如,”載入公司目錄 頁” 。 子方法212、214、216與呼叫者12互動、以決 定特定的回應或動作,如提供呼叫者1 2之語音2 0 8或還 有其它語音所要的資訊或執行所要的動作。 . 經濟部智慧財產局員工消費合作社印製 在階段2 1 8,引擎系統3 0根據子方法2 1 2、 2 1 4、2 1 6的決定提供適當的回應或執行適當的動作。 現請參閱圖1、3、5及6,在安排呼叫線路法2 1 2 的階段2 2 0,安排呼叫線路的頁呈現給呼叫者1 2。引擎 系統3 0播放提示給呼叫者1 2,以指示與安排呼叫線路頁 及從安排呼叫線路頁2 2 0鏈結到其它相關資訊的頁。引擎 系統3 0播放人貝目錄提不給呼叫者1 2,說出呼叫者1 2 -49- 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) 474090 Λ7 -----B7 五、發明說明(47) 想要說話之人的姓名或部門.。I V R系統1 6接收呼叫者回 應提示的語音2 0 8。 (請先閱讀背面之注意事11|:填寫本頁) 在階段2 2 2,引擎系統3 〇從組態與記錄系統3 4得 到安排呼叫線路的字彙,此資訊可以在階段2 2 0之前、之 後或期間得到。在本例中,安排呼叫線路的字彙包括與人員 目錄有關的資料。不過,在其它例中也有其它資料,例如與 航空公司班機時刻系統有關的資訊。 在階段2 2 4,引擎系統3 0決定N個字串可能與呼叫 者1 2所要的字/片語對應。引擎系統3 Q將這:些N —最佳 字串(N -最佳表)與在階段2 2 2所得到的安排呼叫線路 字彙比較。例如,如果N -最佳表中最-信賴之字串的信賴 分數超過上限門檻,則視該字被識別並被接受。信賴分數低 於下限門檻的低-信賴字串不被接受。信賴分數在下限及上 限門檻間的字串被詢問以敗除岐義,如果多個字串的信賴分 數超過上限門檻也是相同情況。 、 經濟部智慧財產局員工消費合作社印製 爲有助於對呼叫者1 2所說字串的識別是唯一的,如有 需要,引擎系統3 0在階段2 2 5執行消除岐義。例如:如 果有兩名員工都是呼叫者1 2所說的姓名,引擎系統3 0嘗. 試經由呼叫者1 2確認他所要之員工所在的部門以區別他們 。因此’引擎系統3 0提示呼叫者1 2 ”說出你想聯絡之人 的部門名稱”。視呼叫者對消除岐義提示的回應而定,引擎 系統3 0選擇其中一個N —最佳字串做爲呼叫者1 2所說的 字串。 在階段2 2 6 ’引擎系統3 0根據在階段2 2 4之語音_ -50- 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公爱) 474090 A7 B7 五、發明說明(48) 與安排呼叫線路字彙的比較決定適當的回應動作。此項決定 可將呼叫安排到所確認的人,或執行請求的動作。 在階段2 2 8,呼叫者的呼叫被安排到識別呼叫者之語 音所確認的人。引擎系統3 〇使用與在階段2 2 6所確認之 人相關的安排呼叫線路資訊,例如分機號碼’將呼叫者1 2 連接到他所想要與其說話的人。例如,如果語音是” John Doe",呼叫者12被安排到:iohn Doe的分機,或另一語音頁,或 客服員,例如,如果語音是”班機時刻表”,則呼叫者1 2 會被安排到飛機排班的客服員。 在階段2 3 0,引擎系統3 0按語音的指示執行非安排 呼叫線路的動作,例如播放或傳真資訊。 現請爹閱圖1、3、5及7 ’圖7顯不資訊擺取法 2 1 4。以下將描述從公司資訊頁得到資訊。本例並非意欲 限制;也可以擷取其它可能_的資訊,包括從其它頁擷取資訊 ,也在本發明的範圍。在階段2 3 2,資訊擷取頁呈現4合呼 叫者1 2。引擎系統3 0播放”載入公司資訊頁”的提示, 當此頁被載入時,再播放”公司資訊頁”的提示。在這些提 示之後,引擎系統3 0播放公司資訊頁上之指示資訊的提示. ,如鏈結到其它語音頁,以及一般的公司資訊。此一般的資 訊包括公司槪況,包括公司的技術以及公司提供的產品及/ 或服務。 在階段2 3 4,引擎系統3 0得到擷取資訊的字彙,用 以識別呼叫者的語音。引擎系統3 0從組態與記錄系統3 4 得到此資訊。此資訊是根據包含在公司資訊頁中的資訊,且_ 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公爱) (請先閱讀背面之注意事填寫本頁) •i -裝 經濟部智慧財產局員工消費合作社印製 -51 - 474090 A7 B7 五、發明說明(49) 這些頁被確認是鏈結自公司資訊頁。引擎系統3 0播放一提 示’例如”你說出以下頁名:”公司歷史”、”新聞及新聞 稿”、”目前事件”,即可鏈結到具有上述資訊的頁。” 在階段2 3 6,引擎系統3 0將呼叫者回應提示所說的 N -最佳字串與擷取資訊的字彙匹配。引擎系統3 0發展出 可能代表呼叫者1 2所說的數個字串。這些字串的N -最佳 被引擎系統3 0選擇與擷取資訊的字彙比較,以決定那一個 字串是呼叫者12所說的字串。 爲有助於唯一識別呼叫者1 2所說的字串,在階段 2 3 8,引擎系統3 0可執行消除岐義。引擎系統3 0可播 放適當的提示給呼叫者1 2,如”我認爲你所說的是”公司 歷’史”不是”目前事件”,對嗎? ”視呼叫者的回應而定, 引擎系統3 0在N -最佳字串中選擇其中之一做爲呼叫者 1 2所說的字串。 _ 在階段2 4 0,引擎系統3 0擷取呼叫者1 2所請、求的 資源。回應在階段2 3 6及2 3 8所識別的唯一字串,引擎 系統3 0使用來自被識別之字串的資訊存取組態與記錄系統 3 4,以擷取與呼叫者之請求有關的資訊。例如,如果呼叫. 者1 2以”是”回應上述消除岐義的問題’則引擎系統3 0 將擷取例如儲存在組態與記錄系統3 4中與公司歷史有關的 公司歷史語音頁。引擎系統3 0播放”載入公司歷史頁”的 提示。 在階段2 4 2,引擎系統3 0將所請求的資源傳送給呼 叫者1 2。在本例中’引擎系統3 0提示呼叫者1 2與公司一 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公嫠) (請先閱讀背面之注意事fPH:填寫本頁) »i 經濟部智慧財產局員工消費合作社印製 -52- 474090 A7 Β7 五、發明說明(5〇) 歷史頁有關的資訊。例如提示可以是”公司歷史頁。你可鏈 結到以下的語音頁...”。 (請先閱讀背面之注意事填寫本頁) 現請參閱圖1、3、5、8 ’圖8顯示交易處理方法 2 1 6。以下將使用訂機票的例子描述。本例並無限制之意 ,也可以處理其它的交易,包括購買產品或商品,都在本發 明的範圍。在階段2 4 4,訂機位的頁呈現給呼叫者1 2。 引擎系統3 0播放”載入訂機位頁”的提示,以及當此頁被 載入時播放另一提示”訂機位頁”。接下來的這些提示,引 擎系統3 0播放訂機位頁上的指示資訊,如鏈結其它語音 頁,其內也包含的一般訂位資訊。此一般資訊包括票價、飛 行中的服務,及/或旅行手續,如報到時間及行李限制。在 階段2 4 6,引擎系統3 0得到識別呼叫者語音的訂位字彙 。引擎系統3 0從組態與記錄系統3 4得到此資訊。此資訊 是根據包含在公司訂機位頁_中的資訊,以及,這些被確認的 頁是鍵結自公司資訊頁。引擎系統3 0播放一提不,例、如” 說出國內航線,國際航線你即可鏈結到具有上述資訊的頁。 Μ * 經濟部智慧財產局員工消費合作社印製 在階段2 4 8,引擎系統3 0將Ν —最佳字串與訂位字 彙匹配。引擎系統3 0發展數個代表呼叫者1 2所說的字串 。這些Ν —最佳字串被引擎系統3 0選擇與訂位字彙比較, 以決定那一個字串是呼叫者12所說的字串。 爲有助於唯一識別呼叫者1 2所說的字串,在階段 2 5 0,引擎系統3 0可執行消除岐義。引擎系統3 0可播 放適當的提示給呼叫者1 2,如”如果你說的是”西北”, 53 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) 474090 A7 __ B7 五、發明說明(51) 請說” 1 ” ,如果你說的是”西南”,請說” 2 ” ,否則說 ”以上皆非”。視呼叫者的回應而定,引擎系統3 0在N — (請先閱讀背面之注意事iH:填寫本頁) 最佳字串中選擇其中之一做爲呼叫者1 2所說的字串。 在階段2 5 2,回應所識別的呼叫者1 2請求,引擎系 統3 0產生一或多個交易請求。回應在階段2 4 8及2 5 0 所決定之唯一識別的字串,引擎系統3 0使用來自所識別之 字串的資訊存取組態與記錄系統3 4,爲與呼叫者之請求有 關的交易擷取資訊。交易請求將起始所請求的交易,或指令 硬體及/或軟體執行所請求的交易。交易請求可:以擷取自組 態與記錄系統3 4,及/或由客戶在擷取自組態與記錄系統 3 4的資訊中插入變數的値以產生交易請求,或完全由客戶 建立的請求產生。例如,如果呼叫者1 2對上述消除岐義之 問題的回應是” 1 ” ,則引擎系統3 0將產生與西北航空有 關的交易請求,例如”訂一張華盛頓D · C .到底特律的往 返機票,3月1日上午8時前往,回程是3月2日下午、10 時離開”。 經濟部智慧財產局員工消費合作社印製 在階段2 5 4,引擎系統3 0將交易請求傳送到引擊系 統3 0中適當的部分,或其它適當的位置,如西北航空的網. 站伺服器。在本例中,引擎系統3 0按照交易請求執行交易 。另者,此項執行的動作可以包括諸如傳送股票的交易單, 或經由傳真伺服器9 7將請求傳真到呼叫者的傳真機2 8, 視呼叫者的請求而定。 在階段2 5 8,引擎系統3 0產生執行交易的回應。在 此,回應指示訂票成功或失敗,如果成功,相關的資訊包括_ -54- 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公蹵) 474030 A7 B7 五、發明說明(52 ) 班機編號、時間、座位及票價等。另者,回應也可指示賣出 或買進股票的價格,或嘗試傳真成功或失敗的資訊給呼叫者 1 2,如果這些交易被請求。 在階段2 6 0,引擎系統3 0提示呼叫者1 2相關的回 應。例如,引擎系統3 0播放的提示例如”你所訂購的機票 是123號班機,3月1日上午8 : 12從華盛頓D · C , 起飛,10 : 48抵達底特律;以及,下午9 : 47的 456號班機,3月3日上午12 : 13抵達華盛頓 D · C . ”或”所請求的資訊己成功地傳真到(‘、6 1 7 ) 5 5 5 - 1 2 1 2, 如有需要,呼叫者1 2回到交易處理頁2 4 4,以便進 行另一次交易。 現請參閱圖1、3、5及9,圖9顯示的方法270是 用於報告及分析互動交談'在階段2 7 2,呼叫者1 2與系 統1 0的其它部分互動交談。來自此交談的資料,包括、呼叫 者12的發言及/或系統1〇所採行的動作被儲存/記錄。 儲存/記錄可在交談期間或之後發生。 ‘ 在階段2 7 4,來自互動交談的儲存資料被監視及/或 報告。報告可以是警告或警報的形式,或是組織資料後的正 式報告供分析。警告可凸顯系統中錯誤的潛在肇因,或至少 系統1 0中要改善的區域。報告可以顯示系統1 〇的性能。 組織性能特徵報告有助於使分析容易,特別是系統1 〇中可 修正的特徵。性能特徵也被組織以便於提供性能報告給 I V R的客戶,以指示客戶所購買的操作良好。 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) (請先閱讀背面之注意古 1 ^--- ':填寫本頁} 經濟部智慧財產局員工消費合作社印製 55 474090 A7SpeechSite ™ IVR systems such as "Previous", "Next" and "Home", as well as other features such as "Find", "Fax" and "Where am I?" 〇In Phase 2 06, the caller Speak Voice 2 0 8 to Phone 2 6 to navigate the entire voice page. Voice 2 0 8 can be a response to the prompts played by the I VR system: system 16, such as requesting a specified voice page, or requesting information or other actions in a commonly spoken language. Speech 208 represents speech-related information, but not necessarily analog or digital speech. For example, speech 2 0 8 may be a set of N-best strings output by the recognizer 3 2. Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs. At stage 2 0, the engine 30 identifies the available sub-methods to determine which sub-method is suitable for processing the request indicated by voice 2 0 8. To identify and identify sub-methods (each of which contains one or more voice modules), the engine system 30 compares the voice 2 0 8 with the sub-method title prompting the caller 12 and / or multiple words. These words Each is related to at least one of the available sub-methods. In a later example, the vocabulary contains synonyms for the title presented to caller 12. For example, if caller 12 says "direction to Boston", method 2000 will advance to stage 2 1 4 to retrieve the call information from the contact us page. If the speech 2 0 8 matches the sub method title (for example, if the speech 2 0 is "company information"), the engine system 3 0 directs the appropriate corresponding sub method to process the speech 2 0 8, such as in the arrangement of the phase 2 1 2 The size of the book is applicable to the Chinese National Standard (CNS) A4 specification (21G X 297 public love) --- 474090 A7 _ B7 V. Description of the invention (46) Calling the line, the information is retrieved at stage 2 1 4 and / Or process transactions in stages 2 1 6. (Please read the Zhuyin on the back? Fill in this page first.) The following will explain the processing of each voice 2 0 8 by each sub-method 2 1 2, 2 14, 2 and 16 in more detail. Play the appropriate prompt to caller 12 for caller 12 to instruct which submethod to direct. For example, if the caller 12 is going to the schedule call line submethod 2 1 2, the engine system 3 will prompt “transfer to the company directory page” (or “transfer to the personnel directory page” or “transfer to the schedule call line page”) . If the caller 12 is to be transferred to the sub method 2 1 4 or 2 1 6 then the prompt played to the caller 1 2 is divided into "Transfer to [Information Retrieval] page" and "Transfer to [Transaction Processing] page ". In this example, "Contact Us" or "Company Information" replaces [Information Retrieval], and "Buy Stock" replaces [Transaction Processing]. In addition, the prompt played to the caller 12 also includes an indication that the appropriate page has been loaded, for example, "Load Company Directory Page". Sub-methods 212, 214, 216 interact with caller 12 to determine a specific response or action, such as providing caller 12 voice 2 08 or other voice-required information or performing the desired action. Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs At stage 2 18, the engine system 30 provides appropriate responses or performs appropriate actions based on the decisions of sub-methods 2 1 2, 2 1 4 and 2 16. Referring now to FIGS. 1, 3, 5 and 6, at the stage 2 2 0 of the call line arrangement method 2 1 2, the page of the call line arrangement is presented to the caller 12. The engine system 30 plays prompts to the caller 12 to indicate and arrange the call line page and links from the call line page 2 2 0 to other related information pages. The engine system 3 0 broadcasts the person's shell catalog to the caller 1 2 and speaks the caller 1 2 -49- This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) 474090 Λ7 ---- -B7 V. Invention Description (47) Name or department of the person who wants to speak. The I V R system 16 receives the caller's response voice 2 0 8. (Please read the note on the back 11 |: Fill in this page first) At stage 2 2 2, the engine system 3 〇 get the vocabulary to arrange the call line from the configuration and recording system 3 4, this information can be before stage 2 2 0, Get after or during. In this example, the vocabulary for arranging call lines includes information related to the directory of people. However, there are other data in other examples, such as information related to airline flight schedule systems. At stage 2 2 4, the engine system 30 determines that N strings may correspond to the words / phrases required by caller 12. Engine system 3Q compares this: some N — best strings (N — best table) with the arranged call line vocabulary obtained in phase 2 2 2. For example, if the trust score of the most-trusted string in the N-best table exceeds the upper threshold, the word is considered to be recognized and accepted. Low-trust strings with trust scores below the lower threshold are not accepted. Strings with trust scores between the lower and upper thresholds are queried to disambiguate. The same is true if the trust scores of multiple strings exceed the upper threshold. Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs To facilitate the identification of the strings mentioned by callers 12 and 2, if necessary, engine system 30 performs disambiguation at stage 2 2 5. For example: if there are two employees with the same name as the caller 12, the engine system 30 will try to identify the department of the employee he wants by the caller 12 to distinguish them. Therefore, the 'engine system 30 prompts the caller 1 2 to "say the department name of the person you want to contact". Depending on the caller's response to the disambiguation prompt, the engine system 30 selects one of the N-best strings as the string called by the caller 12. In the stage 2 2 6 'Engine system 3 0 According to the voice in the stage 2 2 4 _ -50- This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 public love) 474090 A7 B7 V. Description of the invention (48 ) Compared with arranging call line vocabulary to determine the appropriate response. This decision can place the call to the identified person or perform the requested action. At stage 2 2 8 the caller's call is arranged to identify the person identified by the caller's voice. The engine system 30 uses the arrangement call line information related to the person identified in phase 2 2 6 such as extension number 'to connect the caller 1 2 to the person with whom he wants to speak. For example, if the voice is "John Doe", caller 12 is scheduled to: iohn Doe's extension, or another voice page, or customer service, for example, if the voice is "Flight Schedule", caller 1 2 will be Customer service staff scheduled to the aircraft. At stage 230, the engine system 30 performs unscheduled call line actions, such as playing or faxing information, according to the voice instructions. Please refer to Figures 1, 3, 5 and 7 ' Figure 7 shows the method of information display 2 1 4. The following will describe the information obtained from the company information page. This example is not intended to be limited; other possible information can also be retrieved, including information retrieved from other pages. Scope. At stage 2 3 2, the information retrieval page presents the 4th caller 1 2. The engine system 30 plays the prompt of "Loading the company information page", and when this page is loaded, it plays the "company information page" After these prompts, the engine system 30 plays the prompts of the instruction information on the company information page, such as links to other voice pages, and general company information. This general information includes company information, including the company ’s Technology Products and / or services provided by the company. At stage 2 34, the engine system 30 obtains the vocabulary of the captured information to identify the caller's voice. The engine system 30 obtains this information from the configuration and recording system 34. This information is based on the information contained in the company information page, and _ This paper size is applicable to the Chinese National Standard (CNS) A4 specification (210 X 297 public love) (Please read the notes on the back first and fill in this page) • i-装Printed by the Intellectual Property Bureau Employees Cooperative of the Ministry of Economic Affairs-51-474090 A7 B7 V. Description of the invention (49) These pages were confirmed to be linked from the company information page. Engine system 3 0 Play a prompt 'for example, "You say the following page Name: "Company History", "News and Press Releases", "Current Events", you can link to the page with the above information. In phase 2 3 6, engine system 30 matches the caller's response prompt N-best string with the vocabulary of the retrieved information. Engine system 30 develops several that may represent caller 12 Strings. The N-best of these strings is selected by the engine system 3 and compared with the vocabulary of the retrieved information to determine which string is what the caller 12 said. To help uniquely identify the caller 1 The strings referred to by 2, in stage 2 38, engine system 30 can perform disambiguation. Engine system 30 can play appropriate prompts to caller 12 as "I think you are talking about" company Calendar 'history' is not 'current event', right? "Depending on the caller's response, the engine system 30 selects one of the N-best strings as the string said by the caller 12. _ At the stage 2 40, the engine system 30 fetches The resources requested and requested by the caller 12. In response to the unique strings identified in stages 2 3 6 and 2 3 8, the engine system 30 uses the information from the identified strings to access the configuration and recording system 3 4 To retrieve information related to the caller's request. For example, if the caller 1 responds to the above disambiguation question with "yes", then the engine system 3 0 will retrieve, for example, the configuration and recording system 3 The company history voice page related to company history in 4. The engine system 30 plays the prompt of "load company history page". At stage 2 4 2, the engine system 30 transmits the requested resource to the caller 1 2. In In this example, 'Engine system 3 0 reminds callers 1 2 and the company's paper size to apply Chinese National Standard (CNS) A4 specifications (210 X 297 cm) (Please read the note on the back fPH: Fill out this page) » i Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs-52- 474090 A7 Β7 V. Description of the Invention (50) Information about the history page. For example, the prompt may be "company history page." You can link to the following voice page ... "(Please read the notes on the back to complete this page) Now please refer to Figures 1, 3, 5, 8 'Figure 8 shows the transaction processing method 2 1 6. The following will be The example of booking a ticket is described. This example is not meant to be limiting, and other transactions, including the purchase of products or goods, are also within the scope of the present invention. At stages 2 4 4 the page of the booking seat is presented to the caller 1 2. The engine system 30 plays the prompt of “Loading the booking page”, and plays another prompt “booking page” when this page is loaded. Following these tips, the engine system 30 plays the booking Instructions on the flight page, such as links to other voice pages, also contain general reservation information. This general information includes fares, in-flight services, and / or travel procedures such as check-in times and baggage restrictions. At stage 2 4 6 the engine system 30 obtains the reservation vocabulary that recognizes the caller's voice. The engine system 30 obtains this information from the configuration and recording system 34. This information is based on the information contained in the company reservation page_ Information, and these confirmed pages are linked from Company information page. The engine system 30 will not mention it. For example, if you say domestic routes, international routes, you can link to the page with the above information. Μ * Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs At stage 2 4.8, the engine system 30 matches N — the best string with the reservation vocabulary. The engine system 30 develops several strings that represent the callers 12. These N-best strings are selected by the engine system 30 to compare with the reservation word to determine which string is the string said by the caller 12. To help uniquely identify the string called by the caller 12, the engine system 30 may perform disambiguation at stage 250. The engine system 30 can play appropriate prompts to the caller 1 2 such as "if you say" Northwest ", 53 this paper size applies to China National Standard (CNS) A4 (210 X 297 mm) 474090 A7 __ B7 5. Description of the invention (51) Please say "1", if you say "Southwest", please say "2", otherwise say "None of the above". Depending on the caller's response, the engine system 3 0 at N — (Please read the note on the back iH first: fill in this page) Choose one of the best strings as the string called by caller 1 2. At stage 2 5 2, respond to the identified caller 1 2 Request, the engine system 30 generates one or more transaction requests. In response to the uniquely identified strings determined at stages 2 48 and 2 50, the engine system 30 uses the information from the identified strings to access the configuration With the recording system 34, information is retrieved for the transaction related to the caller's request. The transaction request will initiate the requested transaction or instruct the hardware and / or software to execute the requested transaction. The transaction request can be: Self-configuration and recording system 3 4 and / or Variables are inserted into the information of the recording system 34 to generate a transaction request, or are generated entirely by a customer-created request. For example, if the caller 12's response to the above-mentioned disambiguation question is "1", the engine system 3 0 A transaction request related to Northwest Airlines will be generated, such as "Book a round trip ticket to Washington D.C. Detroit, depart at 8 am on March 1 and return at 10 pm on March 2 and leave at 10 am." Printed by the Intellectual Property Bureau's Consumer Cooperatives at stage 254, the engine system 30 transmits the transaction request to the appropriate part of the strike system 30, or other appropriate location, such as Northwest Airlines' web site station server. In this example, the engine system 30 executes the transaction in accordance with the transaction request. In addition, the action performed by this item may include, for example, transmitting a stock transaction order, or faxing the request to the caller's fax machine 2 8 via the fax server 97, Depends on the caller's request. At stage 2 5.8, the engine system 30 generates a response to execute the transaction. Here, the response indicates the success or failure of the booking. If successful, the relevant information includes _- 54- This paper size is in accordance with Chinese National Standard (CNS) A4 (210 X 297 gong) 474030 A7 B7 V. Description of the invention (52) Flight number, time, seat and fare, etc. In addition, the response can also be instructed to sell The price of the stock to buy or buy, or an attempt to fax success or failure information to the caller 12 if these transactions are requested. At stage 2 60, the engine system 30 prompts the caller 12 with a relevant response. For example, the engine A prompt played by system 30 eg "Your flight ticket is Flight 123, departing from Washington DC at 8:12 AM on March 1 and arriving in Detroit at 10:48; and flight 456 at 9:47 PM , March 3rd at 12:13 AM in Washington D.C. ”or“ The requested information was successfully faxed to (', 6 1 7) 5 5 5-1 2 1 2 and caller 1 if needed 2 Go back to the transaction processing page 2 4 4 for another transaction. Referring now to FIGS. 1, 3, 5 and 9, the method 270 shown in FIG. 9 is used to report and analyze interactive conversations' At stage 2 7 2, caller 12 interacts with other parts of system 10 interactively. Data from this conversation, including the caller 12's speech and / or actions taken by the system 10 are stored / recorded. Storage / recording can occur during or after the conversation. ‘At stages 2 7 4, stored data from interactive conversations are monitored and / or reported. Reports can be in the form of warnings or alerts, or a formal report with the data organized for analysis. Warnings can highlight potential causes of errors in the system, or at least areas in the system to be improved. The report can show the performance of the system. Organizational performance characteristics reports help make analysis easier, especially those that can be corrected in System 10. Performance characteristics are also organized to provide performance reports to IVR customers to indicate that the operations they have purchased are good. This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) (Please read the note on the back 1 ^ --- ': Fill in this page} Printed by the Intellectual Property Bureau of the Ministry of Economic Affairs, Consumer Cooperatives 55 474090 A7
(請先閱讀背面之注咅?事填寫本頁) 在階段2 7 6,監視/報告資料被分析。在架構實體或 其它分析實體的人,檢查有關的性能特徵/統計的報告及/ 或警告。例如,他可分析特徵以決定是否有太多要被安排到 員工的呼叫失敗,或是有太多呼叫被安排到客服員或被切斷 。他可以比較識別字彙造成呼叫者1 2連接到員工失敗之發 言的轉錄,以決定〇〇V發言。在階段2 7 6可以執行各種 的分析,以上的分析只是舉例。 從階段2 7 6的分析他可以決定何種修正動作可以及/ 或必須採取。例如,他可決定需要在識別字彙中增加其它的 發音,或者,某人員的姓名或交易的標題有錯誤,不要加到 正確的識別字彙中,以降低〇〇V s。此外,他也可以決定 應在互動交談方法2 0 0中的一或多個部分加入消除岐義特 徵,以降低誤導呼叫的頻率。修正.動作可以使用R C S 3 6加入、刪除I V R 1 6.的警報資訊、提示、鏈結、組態 參數等,以有助於改善系統1 0的操作。在階段2 7 6、所決 定的修正動作在階段2 7 8執行。 經濟部智慧財產局員工消費合作社印製 其它的實施例也都在申請專利範圍的精神與範圍內丨例 如,A/R服務2 4或它的一或多個部分可以提供在I VR . 系統1 6的所在地或在其內部。此外,系統1 0中某些部分 也可與上述的架構不同。例如,也可以在不是Artlsfot®5() Visual Voice Enterprise 的環境中使用。 此外,也可以使用不同的方法分析性能資料。例如,可 以分析相同發言之〇〇V s發生的頻率,同時忽略不常見的 〇〇V發言。具有相同特徵的〇〇V發言可以群組在一起,_ -56- 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公« ) 474030 A7 __ B7 五、發明說明(54) (請先閱讀背面之注意事填寫本頁) 如此,他可以從群組中聽到足夠發言以確認0 0 V發言。完 成此的方法是蒐集所有識別失敗或識別信賴度低的發言波形 (u 1 a w s的形式)。使用標準的識別器前端將每一個 ulaw轉換成一序列的特徵向量(例如Mel-Frequency Cersmal Coefficients(MFCCs ))。爲每一個語音框(例如 1 毫秒)產生一個M F C C向量。使用特徵向量的動態排列將 相同的發言集束在一起,或是使用諸如k-means的集束技術 。每一個叢集代表0 0V之發言的集合加上一些雜訊。人操 作的轉錄器傾聽叢集中少許的發言以從叢集中決定主要的 〇〇V。集束有助於轉錄器不需聽所有的發言才能識別 〇〇V 。 此外,也可以使用自動技術來轉錄發言。例如,也可由 電話環路識別器產生語音的陳述來轉錄發言,以取代人的轉 錄。可以使用此種方法從银一個發言的叢集中轉錄少許的發 言。語音的陳述可與語音的字典交叉參考,或傳送給人、·以確 認〇0 V。0〇V發言可被標示以考慮修正動作。另者,發 言也可與較大的字典比較,例如名稱。 * 經濟部智慧財產局員工消費合作社印製 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公爱) -57-(Please read the note on the back first? Fill in this page first) At stage 2 7 6 the monitoring / reporting data is analyzed. Someone at the construction entity or other analysis entity checks reports and / or warnings about performance characteristics / statistics. For example, he can analyze the characteristics to determine if too many calls to be scheduled to an employee fail, or if too many calls are scheduled to an agent or disconnected. He can compare the transcription of the recognition vocabulary that caused the caller 12 to connect to the employee's failed speech to determine the 00V speech. Various analyses can be performed at stages 2 7 6. The above analysis is only an example. From the analysis of stages 2 7 6 he can decide what corrective actions can and / or must be taken. For example, he may decide that other pronunciations need to be added to the recognition vocabulary, or that the name of a person or the title of a transaction is wrong, and should not be added to the correct recognition vocabulary to reduce 00V s. In addition, he may decide to include disambiguation features in one or more of the interactive conversation methods 200 to reduce the frequency of misleading calls. Correction. Actions can use R C S 3 6 to add and delete I V R 16 6. Alarm information, prompts, links, configuration parameters, etc., to help improve the operation of the system 10. In phase 2 7 6, the determined corrective action is performed in phase 2 7 8. Other embodiments printed by the Intellectual Property Bureau of the Ministry of Economic Affairs and the Consumer Cooperatives are also within the spirit and scope of the scope of patent application. For example, A / R service 2 4 or one or more parts of it can be provided in I VR. System 1 6 is located in or within it. In addition, some parts of the system 10 may be different from the above-mentioned architecture. For example, it can also be used in environments other than Artlsfot®5 () Visual Voice Enterprise. In addition, different methods can be used to analyze performance data. For example, you can analyze the frequency of 00V s for the same speech, while ignoring less common 00V speeches. The 00V speeches with the same characteristics can be grouped together. _ -56- This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 public «) 474030 A7 __ B7 V. Description of the invention (54) (Please First read the notes on the back and fill out this page) So he can hear enough speeches from the group to confirm 0 0 V speeches. The method to accomplish this is to collect all the speech waveforms (in the form of u 1 a w s) that have failed recognition or have low recognition reliability. Use a standard recognizer front end to convert each ulaw into a sequence of feature vectors (such as Mel-Frequency Cersmal Coefficients (MFCCs)). Generate an M F C C vector for each speech frame (for example, 1 millisecond). Use dynamic permutation of feature vectors to cluster the same speech together, or use clustering techniques such as k-means. Each cluster represents a set of speeches of 0 0V plus some noise. A human-operated transcriber listens to a few speeches in the cluster to determine the main 00V from the cluster. Bunching helps the transcriber not need to listen to all speech to recognize 〇〇V. Alternatively, speech can be transcribed using automated techniques. For example, speech can be transcribed by a telephone loop recognizer to replace a person's transcript. This method can be used to transcribe a small number of speeches from Yin Yin's speech cluster. Speech statements can be cross-referenced with a dictionary of speech, or passed to a person to confirm 0 V. 〇V speech can be marked to consider corrective action. Alternatively, speech can be compared to larger dictionaries, such as names. * Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs
Claims (1)
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US14460999P | 1999-07-20 | 1999-07-20 |
Publications (1)
Publication Number | Publication Date |
---|---|
TW474090B true TW474090B (en) | 2002-01-21 |
Family
ID=22509353
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
TW89114557A TW474090B (en) | 1999-07-20 | 2000-07-20 | Speech-enabled information processing |
Country Status (1)
Country | Link |
---|---|
TW (1) | TW474090B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7215743B2 (en) | 2002-12-20 | 2007-05-08 | International Business Machines Corporation | Telephony signals containing an IVR decision tree |
TWI841866B (en) * | 2021-09-14 | 2024-05-11 | 中國信託商業銀行股份有限公司 | Business handling willingness determination method and computing device thereof |
-
2000
- 2000-07-20 TW TW89114557A patent/TW474090B/en not_active IP Right Cessation
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7215743B2 (en) | 2002-12-20 | 2007-05-08 | International Business Machines Corporation | Telephony signals containing an IVR decision tree |
TWI841866B (en) * | 2021-09-14 | 2024-05-11 | 中國信託商業銀行股份有限公司 | Business handling willingness determination method and computing device thereof |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US7283973B1 (en) | Multi-modal voice-enabled content access and delivery system | |
US7609829B2 (en) | Multi-platform capable inference engine and universal grammar language adapter for intelligent voice application execution | |
US8718242B2 (en) | Method and apparatus for automatically building conversational systems | |
US8036897B2 (en) | Voice integration platform | |
US9152375B2 (en) | Speech recognition interface for voice actuation of legacy systems | |
US7016843B2 (en) | System method and computer program product for transferring unregistered callers to a registration process | |
KR100459299B1 (en) | Conversational browser and conversational systems | |
US9626959B2 (en) | System and method of supporting adaptive misrecognition in conversational speech | |
EP1481328B1 (en) | User interface and dynamic grammar in a multi-modal synchronization architecture | |
US20050234727A1 (en) | Method and apparatus for adapting a voice extensible markup language-enabled voice system for natural speech recognition and system response | |
US20030115056A1 (en) | Employing speech recognition and key words to improve customer service | |
EP2113909B1 (en) | Location-based services | |
JP2007527640A (en) | An action adaptation engine for identifying action characteristics of a caller interacting with a VXML compliant voice application | |
US20120065982A1 (en) | Dynamically generating a vocal help prompt in a multimodal application | |
US20050091057A1 (en) | Voice application development methodology | |
US20090304161A1 (en) | system and method utilizing voice search to locate a product in stores from a phone | |
US20020169604A1 (en) | System, method and computer program product for genre-based grammars and acoustic models in a speech recognition framework | |
US20020188443A1 (en) | System, method and computer program product for comprehensive playback using a vocal player | |
US20030055651A1 (en) | System, method and computer program product for extended element types to enhance operational characteristics in a voice portal | |
US20060069570A1 (en) | System and method for defining and executing distributed multi-channel self-service applications | |
US20030149565A1 (en) | System, method and computer program product for spelling fallback during large-scale speech recognition | |
TW474090B (en) | Speech-enabled information processing | |
CA2379853A1 (en) | Speech-enabled information processing | |
US20020099545A1 (en) | System, method and computer program product for damage control during large-scale address speech recognition | |
US20020099544A1 (en) | System, method and computer program product for damage control during large-scale address speech recognition |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
GD4A | Issue of patent certificate for granted invention patent | ||
MM4A | Annulment or lapse of patent due to non-payment of fees |