JPH09319297A - Communicating device by speech and writing - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a communicating device by speech and writing which enables a deaf-mute and a person with a normal, healthy body to easily make conversions. SOLUTION: This device has a pen 11 for character input, a table 1 for character input, a display panel 2, a microphone 3 for speech input and a speaker 4 for speech output and further a character and speech recognition/output control means consisting of the character recognizing section 51, speech recognizing section 53, rule speech synthesizing section 52, display section 54, etc., controlled by a control section 55. The deaf-mute writes the contents desired to be transmitted by himself on the tablet 1 for character input by means of the pen 11 for character input. The written characters are encoded by the character recognizing section 51 and the codes are converted to speech output by the rule speech synthesizing section 52. The speech output is transmitted as a speech from the speaker 4 for speech output to the person with the normal, healthy body. The person with the normal, healthy body speaks toward the microphone 3 for speech input. The speech input is encoded in the speech recognizing section 53 and is converted to character data in the display section 54. The character data is displayed as a character on the display panel 2. The deaf-mute recognizes the will of the person with the normal, healthy body by viewing the display panel 2. As a result, the smooth conversations are made possible.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is effective when applied to a voice writing device, and more particularly to a voice writing device as a welfare device that allows a deaf person and a healthy person to freely talk using character recognition and voice recognition. Regarding technology.

[0002]

2. Description of the Related Art Conventionally, when a deaf person conveys a will to a healthy person or has a conversation, he or she uses sign language or writing.

On the other hand, Ohmic's "Microcomputer Handbook", published December 15, 1985, P391-P393,
It describes a character recognition technique for reading characters. Also, in the same magazines P903 to P924, voice system techniques are described, and in P916 and P917, a technique (voice synthesis from text) for synthesizing voice from data (text data) based on a character sequence written according to a normal writing method. , P918 and onward describes voice recognition.

The Institute of Electrical Engineers of Japan, Journal C, Vol. 108, No. 10, P85
0 to P857 describe voice recognition technology, recording / playback technology, and the like.

[0005]

DISCLOSURE OF THE INVENTION Since conversations between deaf people who have acquired sign language and between deaf people and normal people are in sign language, they can be conducted relatively smoothly. However, the learning rate of sign language in healthy people is low, and the conversation between a healthy person who is not aware of sign language and the deaf person is gesturing and gesturing, which makes it difficult to convey an accurate will and entails great difficulty.

[0006] An object of the present invention is to provide a voice writing instrument capable of accurately transmitting a will between a deaf person and a normal person by a simple operation.

The above and other objects and novel features of the present invention will be apparent from the description of this specification and the accompanying drawings.

[0008]

The following is a brief description of an outline of typical inventions disclosed in the present application.

(1) Character input means, voice input means,
Character recognition which has a display means, a voice output means, and a character voice recognition / output control means, wherein the character voice recognition / output control means recognizes a character input by the character input means and outputs character data. Section, a regular voice synthesizing section for converting to voice data based on the character data of the character recognizing section and outputting the voice data to the voice output means, and a voice input by the voice input means for recognizing the voice data and outputting the voice data. The voice recognition unit includes a voice recognition unit, a display unit which converts the voice data of the voice recognition unit into character data and outputs the character data to the display unit, and a control unit which controls the respective units.

The character recognition unit includes an interface unit, a feature extraction unit that receives character data from the character input unit to extract character features, a character standard pattern memory, character data from the feature extraction unit, and the character standard. The voice recognition unit includes a collation unit that collates a character standard pattern of a pattern memory, and a determination processing unit that determines whether the character data by the feature extraction unit matches the character standard pattern and outputs the matched text data. Is an interface unit, an A / D converter for receiving analog voice data from the voice input unit, a voice feature extraction unit for extracting voice features of the digital voice data converted by the A / D converter, and a voice standard. Pattern memory, voice data by the voice feature extraction unit, and voice standard pattern of the voice standard pattern memory And a determination processing unit that determines whether the voice data by the voice feature extraction unit matches the voice standard pattern and outputs the matched text data. The display unit includes an interface unit and the characters. / A controller for outputting character data to the display means based on the text data output from the voice recognition unit, wherein the regular voice synthesis unit includes an interface unit, a dictionary memory, a sound source memory, the dictionary memory and the A language processing / waveform connection processing unit that converts the text data output from the character recognition unit into a voice waveform based on the information in the sound source memory, and converts the voice data of the language processing / waveform connection processing unit into analog voice data. And a D / A converter for outputting to the audio output means.

The regular voice synthesizing unit is provided with a recording / reproducing unit for recording and reproducing desired voice data based on input data.

The character input means is a touch panel type character input tablet equipped with a character input pen and a keyboard, the display means is a liquid crystal display panel, the voice input means is a voice input microphone, and The voice output means is a voice output speaker, and the character input tablet is superposed on the liquid crystal display panel so that the display surface and the character input surface are flush with each other.

The voice writing device is used as follows. (1) A deaf person inputs a message that he / she wants to convey to a healthy person into a character input tablet with a character input pen. The input character data is recognized by the character recognition unit and then output as text data from the character recognition unit. This text data is processed by the regular voice synthesizing unit, converted into voice data as a corresponding word or sentence, and emitted as voice from the voice output speaker. As a result, the message of the deaf person can be accurately transmitted as voice to the healthy person.

(2) A healthy person speaks a message he or she wants to convey to the deaf person into the voice input microphone. The input voice data is recognized by the voice recognition unit and then output as text data from the voice recognition unit. This text data is processed by the display unit and displayed as a corresponding word or sentence on the display panel. As a result, the message of the healthy person is accurately transmitted to the deaf person as a character.

(3) Further, the deaf person presses the keyboard of the character input tablet, so that the prerecorded sentence is emitted from the voice output speaker and is accurately transmitted to the healthy person. For example, a voice output such as "I want to talk using this machine because I have a disability and cannot hear well, please nod." Is output.

According to the above-mentioned means (1), the deaf person can convey the message of the deaf person to the normal person by voice, simply by inputting characters to the character input tablet using the input pen. The response by the normal person only speaks into the microphone for voice input, and since the message of the normal person is displayed as characters on the display panel, it is not necessary for the normal person to use written language without using sign language. Will be able to have a smooth conversation.

Further, prior to the conversation, the deaf person can convey a desired message to a healthy person by pressing the keyboard of the character input tablet. This message can be sent during or after the conversation, which reduces the burden of the deaf person on the conversation.

[0018]

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In all the drawings for describing the embodiments of the present invention, components having the same functions are denoted by the same reference numerals, and their repeated description will be omitted.

(Embodiment 1) FIGS. 1 to 10 are diagrams relating to a voice writing device according to an embodiment of the present invention. FIG. 1 is a block diagram showing a system configuration of the voice writing device, and FIG. 2 is a voice writing device. 3 is a perspective view showing the external appearance of the device, FIG. 3 is a block diagram showing the system configuration of a character recognition unit, a regular voice synthesis unit, a voice recognition unit and a control unit, FIG. 4 is a block diagram showing the configuration of the character recognition unit, and FIG. FIG. 6 is an explanatory diagram showing a correlation between a direction code for character recognition and an input character, FIG. 6 is an explanatory diagram showing a recognized character recognized and a direction code string thereof, and FIG. 7 is a block diagram showing a configuration of a rule voice synthesis unit. FIG. 8 is an explanatory view showing the principle of regular voice synthesis, FIG. 9 is a block diagram showing the configuration of the voice recognition unit, and FIG. 10 is a flowchart showing how to use the voice writing device.

The voice writing device of this embodiment is a pen-PC type integrated display / input device 5 as shown in FIG.

A character input pen 11 is connected to the display / input integrated device 5 via a cable 7.

Further, the display panel 2 constitutes display means in the inclined upper surface portion of the display / input integrated device 5. Although not shown, the display panel 2 is composed of a liquid crystal display panel. A touch panel type character input tablet 1 is provided on the liquid crystal display panel. This character input tablet 1
Constitute the keyboard 6. Therefore, the display surface and the character input surface have the same surface structure.

The character input pen 11 and the character input tablet 1 constitute a character input means. Therefore, with the character input pen 11, the character input tablet 1
Enter characters by writing on them (coordinate input)
can do. Also, by pressing a predetermined function key of the keyboard 6 with the character input pen 11, the voice writing device can be operated.

On the other hand, a voice input microphone 3 as voice input means is connected to the display / input integrated device 5 via a cable 8. A voice can be input by uttering (speaking) a sound into the voice input microphone 3.

Further, the display / input integrated device 5 has a built-in audio output speaker 4 as audio output means.

On the other hand, in the display / input integrated device 5,
As shown in FIG. 1, a character voice recognition / output control means 15 is built in.

The character voice recognition / output control means 15 recognizes the characters input to the character input tablet 1 and outputs them as text data, and the text data output from the character recognition unit 51. A regular voice synthesizing unit 52 for converting the character data into voice data and outputting the voice data to the voice output speaker 4, and the voice input microphone 3
A voice recognition unit 53 that recognizes a voice input to and outputs it as text data, and a display unit 54 that converts the voice data from the text data output from the voice recognition unit 53 into character data and outputs the character data to the display panel 2. And a control unit 55 for controlling the character recognition unit 51, the regular voice synthesis unit 52, the voice recognition unit 53, and the display unit 54.

Next, the character recognition unit 51, the regular voice synthesis unit 52, the voice recognition unit 53, the display unit 54, and the control unit 55 will be described.

The control unit 55 is composed of a CPU (central processing unit) and, as shown in FIG. 3, has a bus A21 and a bus B22.
And a character recognition unit 51, a regular voice synthesis unit 52, a voice recognition unit 53, and a display unit 54 which are connected to the bus A21 and the bus B22, respectively.

As shown in FIGS. 3 and 4, the character recognition unit 51 includes a determination processing unit 511, a feature extraction unit 512, and a collation unit 5.
13, an interface unit 515 is connected to the bus C23, and the bus A2 is connected via the interface unit 515.
1 and the bus B22. A character standard pattern memory 514 is connected to the collating unit 513.

The character recognition unit 51 constitutes an online character recognition function for recognizing a character when the deaf person writes a message that he / she wants to convey to a normal person with the character input pen 11 on the character input tablet 1.

The feature extracting section 512 is shown in FIG. 3 and FIG.
As shown in (1), the output (coordinate data) from the character input tablet 1 is received and a direction code is created from the coordinate data.

For example, as the direction code, as shown in FIG. 5, the input characters are displayed by eight-direction codes (0 to 7) at 45-degree intervals that are vertically and horizontally inclined. A number or a character is indicated by a direction code string due to its characteristics.
For example, the direction code of the number "2" starts from the start indicated by S for writing the number "2" and goes to the diagonally rightward direction (1), the horizontal rightward direction (0), and the diagonal rightward direction (0) as shown by the arrow. 7), diagonal left downward direction (5), vertical downward direction (6), horizontal right direction (0), and so on until the end of writing E is reached, as shown in FIG. 6, a chain of direction code strings of 107560. It is represented by. The numerical value "2" reproduced by the direction code string of 107560 is displayed only by each direction code component as shown on the right side of FIG.

In the character standard pattern memory 514, each direction code string is previously stored as a standard pattern for each target character.

The collation unit 513 is the feature extraction unit 51.
The direction code sequentially sent from 2 and the direction code in the character standard pattern memory 514 are compared with each other, and the best match is set as the answer.

The collating unit 513 is the 10th journal of the Institute of Electrical Engineers of Japan.
It is composed of a pattern matching circuit using dynamic programming as described in Volume 8, No. 10, 1988, P850-P854.

This pattern matching circuit absorbs the non-linear expansion / contraction of each character when calculating the difference between the input and the standard pattern, and calculates the similarity as a whole. The equations (1) and (2) shown in P853 of the above-mentioned IEEJ Transactions C, Vol. 108, No. 10 are shown below.

[0038]

[Equation 1]

Where d (i ₁ j) is the Chebyshev distance between the input i ₁ and the standard j-th frame feature amount, x _i k is the input i-th frame and k-channel feature amount, and Y _j k is the standard value. j
The feature amount of the k-th channel in the frame, l is the number of channels of the feature amount, where l = 16.

[0040]

[Equation 2]

Here, D (i, j) is a matching value in the input i, standard j frame, and C (i, j) is a value indicating misalignment from the linear matching line in the input i, standard j frame. is there.

Although these equations 1 and 2 show the pattern matching processing used in voice recognition, they can be applied in the same way to character recognition.

Further, in addition, in voice recognition, the input at each time of input and the standard pattern are matched, and in character recognition, only the pattern matching between the directional codes of the input and the standard pattern is different. is there.

The determination processing unit 511 is configured to detect the collation unit 51.
3 is received, a judgment is made, the inputted character is finally decided, and it is outputted as text data. This text data is sent from the interface unit 515 to the bus A21 and the bus B22. The determination processing unit 511 also performs text processing.

In the character recognition unit 51, the input character is converted into a direction code string (input character direction code string) as character data by the feature extraction unit 512. The input character direction code string is sequentially sent to the matching unit 513, and is sequentially matched with the direction code string stored in the character standard pattern memory 514, and the one that matches best is the answer. The determination processing unit 511 finally determines the direction code string used as the answer and outputs it as text data. This text data is sent from the interface unit 515 to the bus A21 and the bus B22.

As shown in FIGS. 3 and 7, the regular voice synthesizing unit 52 connects the language processing / waveform connection processing unit 5 to the bus D24.
21, a dictionary memory 522, a sound source memory 523, a D / A converter 524, an interface unit 525, and a recording / playback unit 526 are connected. The regular voice synthesizing unit 52 constitutes a regular voice synthesizing function for inputting the text data from the character recognizing unit 51 and voice-outputting a corresponding word or sentence.

The dictionary memory 522 stores readings and accents corresponding to words.

The sound source memory 523 is used for the language processing /
The phonemes corresponding to the pronunciations and accents output by the language processing unit of the waveform connection processing unit 521 are stored.

The D / A converter 524 connects the phonemes in the sound source memory 523 according to a command from the language processing / waveform connection processing unit 521 and outputs the audio data as analog audio data. Output to.

The interface section 525 is connected to the bus A21 and the bus B22 as shown in FIG. The text data is received via the interface unit 525.

The recording / playback unit 526 comprises a voice expansion processing unit 5261 connected to the bus D24, and a compressed data memory 5262 connected to the voice expansion processing unit 5261. The voice expansion processing unit 5261 is, for example, the Institute of Electrical Engineers of Japan, Volume 108, No. 10, 1988, P8.
As described in 54-P856, ADPCM (Adapti
ve Differential Pulse Code Modulation) It is configured by compression / expansion method. Compressed data memory 5262
Stores audio data compressed by the ADPCM compression / expansion method.

The language processing / waveform connection processing section 521 comprises a language processing section and a waveform connection processing section. The language processing unit refers to the dictionary memory 522 to refer to the interface unit 5
The text data received via 25 is determined as a word, and the reading and accent corresponding to the word are output.
Further, the waveform connection processing unit reads out the phoneme corresponding to the reading and accent output from the language processing unit from the sound source memory 523 and outputs it.

FIG. 8 shows the principle of determining text data as a word by the language processing / waveform connection processing unit 521 with reference to the dictionary memory 522.

For example, when the text data input is represented by a string of characters "I am Gyujutsusha.", The language processing determines the word (Kanji) as "I am / technologist / is." Is given. After that, waveform connection processing is performed.

As shown in the lower part of FIG. 8, the character "wa",
“Ta” and “shi” are represented as specific waveforms (phonemes). Therefore, the waveform connection processing unit continuously connects the waveforms of the characters "wa", "ta", and "shi". Since this waveform is output to the voice output speaker 4 as analog voice data by the D / A converter 524, the word "I" is uttered from the voice output speaker 4.

In the rule voice synthesis unit 52, the language processing unit of the language processing / waveform connection processing unit 521 refers to the dictionary memory 522 to determine the text data as a word, and based on the reading and accent corresponding to the word. Connect phonemes and output as voice data.

As a result, the characters and words (sentences) input by the deaf person into the character input tablet 1 are output as voice from the voice output speaker 4, and are therefore accurately transmitted to the healthy person.

On the other hand, the voice writing device of the present embodiment outputs a voice (fixed phrase) having a predetermined content by pressing a predetermined function key of the keyboard 6 of the character input tablet 1 when talking to a deaf person. It is configured to do. This phrase is recorded in the recording / playback unit 526 and is played back as necessary. Keyboard 6
By operating, it is possible to newly store a predetermined cliché or to correct and delete it.

For example, as a fixed phrase (talking),
"I have a disability and I can't hear well, so I want to use this machine to talk, but if you don't mind, please nod me."

As shown in FIG. 3 and FIG. 9, the voice recognition unit 53 includes a determination processing unit 531, a voice feature extraction unit 532, a collation unit 533, an interface unit 535, and A / E on the bus E25.
The D converter 536 is connected. In addition, the matching unit 5
A voice standard pattern memory 534 is connected to 33. The interface unit 535 is connected to the bus A21 and the bus B22 as shown in FIG. The voice recognition unit 53 has a voice recognition function of recognizing a voice input by a healthy person via the voice input microphone 3.

The voice recognition unit 53 is the CJ
It is basically the same as that described in Vol. 108, No. 10, P853 to P857.

The A / D converter 536 receives the analog voice data input to the voice output speaker 4,
Convert to digital audio data.

The speech feature extraction unit 532 has the same structure as the one shown in the Institute of Electrical Engineers of Japan C, Vol. 108, No. 10, P85.
It has a filter processing configuration of 6 channels. This one
The 6-channel filtering process is performed to extract the characteristics of each voice from the voice input from the voice input microphone 3. That is, the uttered contents enter the voice recognition unit 53 via the voice input microphone 3 and the voice feature extraction unit 53.
2, for example, bandpass filtering is performed.

The collation unit 533 is the character recognition unit 51.
The matching unit 513 has a matching circuit similar to that of the matching unit 513, and finds the difference between the voice input from the voice input microphone 3 and the voice standard pattern at each time point to obtain the overall similarity, and gives the best match. And That is, the collating unit 533 collates the bandpass filter output group corresponding to each recognition target word stored in the voice standard pattern memory 534 with the bandpass filter output sequentially sent from the voice feature extracting unit 532. .

In the voice standard pattern memory 534,
The filter output (speech standard pattern) corresponding to each word to be recognized is stored.

The determination processing unit 531 is provided with the collation unit 53.
The output of 3 is received and a determination is made, and the one that most matches the input is finally determined as the recognition result. Text data is transferred from the interface unit 515 to the bus A21 and the bus B2.
Sent to 2. The determination processing unit 531 also performs conversation processing.

In the voice recognition unit 53, the voice that has been input and becomes digital voice data is extracted by the voice feature extraction unit 532, and is compared with the voice standard pattern of the voice standard pattern memory 534 by the collation unit 533, and the determination processing unit 531 is performed. Is finally determined in and output as text data. This text data is sent from the interface section 535 to the bus A21 and the bus B22.

Although not shown in the figure, the display section 54 is provided with the bus A.
21 and an interface section connected to the bus B22, and a controller for controlling the liquid crystal display panel forming the display panel 2 is connected to the bus connected to the interface section. This controller is composed of, for example, a liquid crystal timing controller and a data driver. The display unit 54 is the voice recognition unit 53.
The character data is output to the display panel 2 so that the text data from is input and the corresponding word or sentence is displayed. The display unit 54 constitutes a display function for presenting the deaf person.

Next, a method of using the voice writing device of this embodiment will be described with reference to the flowchart of FIG.

When the deaf person wants to have a conversation with a healthy person, the conversation is started (step 101).

The deaf person talks to a healthy person that he / she wants to have a conversation. This narrative is a fixed phrase, and is performed by the deaf person pressing down a specific function key (narrative key) on the keyboard 6 of the character input tablet 1 (step 102).

By pressing down the function key,
The control unit 55 detects this and activates the recording / playback unit 526 of the regular voice synthesis unit 52. As a result, the recording / playback unit 5
The previously stored data is taken out from the compressed data memory 5262 of 26, is decompressed by the audio decompression processing unit 5261 and is returned to the original audio data, and the D / A converter 52
At 4, the analog voice data is converted and output to the voice output speaker 4.

As a result, for example, the voice output from the voice output speaker 4 is, for example, "I have a disability and cannot hear well, so I want to talk using this machine, but nod if you like." Sent out (step 10)
3).

The deaf person, after confirming that the healthy person to be talked with has agreed with the narrative and nodded, writes a desired character or the like on the character input tablet 1 using the character input pen 11 to obtain a normal sound. Tell the will to others. The will to convey depends on the situation such as the route to the station and where the item you want to buy is. For example, a deaf person inputs "Please tell me the route to Tokyo station."

When the characters are input to the character input tablet 1, the character recognizing unit 51 activated by the control unit 55 recognizes them and the regular voice synthesizing unit 52 receives the interface unit 5.
The corresponding text data is transferred via 25 (step 104).

In the rule voice synthesis unit 52, the language processing unit of the language processing / waveform connection processing unit 521 refers to the dictionary memory 522 to determine the text data as a word, and based on the reading and accent corresponding to the word. Output as voice data. This voice data is converted to D / A converter 52.
4 is converted to analog voice data and output to the voice output speaker 4 (step 105).

A healthy person can know the intention of the deaf person accurately by listening to the words emitted from the voice output speaker 4.

Therefore, the healthy person speaks the words for the question of the deaf person into the voice input microphone 3. In response to the above-mentioned question, say, "For example, if you go to the second floor for a second time, there will be a city near you" (step 106).

The spoken contents enter the voice recognition unit 53 via the voice input microphone 3 and are subjected to, for example, band pass filter processing by the voice feature extraction unit 532. At the time of recognition, the collating unit 533 compares the bandpass filter output group corresponding to each word to be recognized, which is stored in advance in the voice standard pattern memory 534, and the bandpass filter output sequentially sent from the voice feature extracting unit 532. The result of the collation is further determined, and the output of the collation is determined by the determination processing unit 531 as the recognition result. The contents recognized by the voice recognition unit 53 are sent as text data from the interface unit 535 (step 107).

The text data sent from the voice recognition unit 53 is input to the controller composed of the liquid crystal timing control unit and the data driving unit of the display unit 54 via the interface unit 535. As a result, the controller drives the display panel 2 which is a liquid crystal display panel to display characters on the display surface of the display panel 2 (step 108).

On the display surface of the display panel 2, "If you go straight on this street and turn right at the second traffic light, you will find Tokyo Station in the immediate vicinity." It will be displayed in a mixture of Kanji characters. .

The deaf person can accurately understand the contents spoken by the healthy person by reading (visually) the characters displayed on the display panel 2, understand the answer to the question, and once the conversation is completed (end). (Step 109).

When the deaf person feels dissatisfied with the content of a normal person's answer or wants to talk about something else, that is,
If the user wants to continue the conversation further, the process returns to step 104, the character is input again to the character input tablet 1, and the conversation is continued by following the above procedure. The loop of steps 104 to 109 is repeatedly executed until it is desired to end the conversation.

When the deaf person wants to end the conversation, the deaf person again depresses the function key of the keyboard 6 of the character input tablet 1 (step 110).

By pressing down the function key, the conversation end message from the recording / reproducing unit 526 in the rule voice synthesizing unit 52, for example, a phrase such as "Thank you very much, I understand very well" etc. (Step 111).

When the healthy person hears and agrees with the conversation end message, the conversation between the deaf person and the normal person is completed (step 112).

The end of the conversation can be naturally expressed by a healthy person.

By using the voice-writing device of the present embodiment, the conversation between the deaf person and the normal person becomes accurate and simple without knowing the sign language.

Further, since a healthy person simply speaks into the voice input microphone and does not need troublesome writing, the burden is reduced and the conversation with the deaf person becomes pleasant.

Although the invention made by the present inventor has been specifically described based on the embodiments, the present invention is not limited to the above embodiments, and various modifications can be made without departing from the scope of the invention. Needless to say. For example, if a structure is provided in which two sets of the character input tablet 1 and the character input pen 11 as the character input means are provided, even a healthy person can input with a figure.

[0091]

The effects obtained by the typical ones of the inventions disclosed in the present application will be briefly described as follows.

(1) If the deaf person writes a message on the character input tablet using handwritten characters that he or she is accustomed to on a daily basis, the corresponding content is output as voice to the other person (healthy person), and the normally person is deaf and deaf. Can know the will of the person accurately. In addition, if a normal person speaks a normal conversation voice into the voice input microphone, the content is displayed as characters on the display panel, so that the deaf person can know the will of the normal person accurately and surely. it can. Therefore, it is possible to accurately convey the intention between the deaf person and the healthy person by a simple operation.

(2) A conversation between a deaf person and a normal person is established because the deaf person writes a character or the like by hand on the display panel and the normal person speaks into the voice input microphone.
You can talk without knowing sign language.

(3) The healthy person can talk with the deaf person by speaking into the voice input microphone without handwriting the intention to the deaf person, and the burden on the normal person is lightened.

[Brief description of drawings]

FIG. 1 is a block diagram showing a system configuration of a voice writing device according to an embodiment of the present invention.

FIG. 2 is a perspective view showing an external appearance of the voice writing device of the present embodiment.

FIG. 3 is a block diagram showing a correlation among a control unit, a character recognition unit, a regular voice synthesis unit, and a voice recognition unit in the system configuration of the voice writing device of the present embodiment.

FIG. 4 is a block diagram showing a configuration of a character recognition unit of the voice writing device of the present embodiment.

FIG. 5 is an explanatory diagram showing a correlation between a direction code for character recognition and an input character in the voice writing device of the present embodiment.

FIG. 6 is an explanatory diagram showing a recognized character recognized by the voice writing device of the present embodiment and a direction code string thereof.

FIG. 7 is a block diagram showing a configuration of a regular voice synthesis unit in the voice writing device of the present embodiment.

FIG. 8 is an explanatory diagram showing the principle of regular voice synthesis in the voice writing device of the present embodiment.

FIG. 9 is a block diagram showing a configuration of a voice recognition unit in the voice writing device of the present embodiment.

FIG. 10 is a flowchart showing a method of using the voice writing device of the present embodiment.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Tablet for character input, 2 ... Display panel, 3 ... Microphone for audio input, 4 ... Speaker for audio output, 5 ... Integrated display / input device, 6 ... Keyboard, 7, 8 ... Cable, 11
... input pen, 15 ... character voice recognition / output control means, 21
... bus A, 22 ... bus B, 23 ... bus C, 24 ... bus D, 25 ... bus E, 51 ... character recognition section, 52 ... regular speech synthesis section, 53 ... speech recognition section, 54 ... display section, 55 ... Control unit, 511 ... Judgment processing unit, 512 ... Feature extraction unit, 513
... collation unit, 514 ... Character standard pattern memory, 515 ...
Interface unit, 521 ... Language processing / waveform connection processing unit, 522 ... Dictionary memory, 523 ... Sound source memory, 524
... D / A converter, 525 ... Interface unit, 526 ...
Recording / playback unit, 5261 ... Voice decompression processing unit, 5262 ... Compressed data memory, 531 ... Judgment processing unit, 532 ... Voice feature extraction unit, 533 ... Collation unit, 534 ... Voice standard pattern memory, 535 ... Interface unit, 536 ... A / D converter.

Continuation of front page (51) Int.Cl. ⁶ Identification number Office reference number FI Technical display location G10L 3/00 551 G10L 5/04 E 5/04 G06F 15/62 321A 322K

Claims (4)

[Claims] 1. A character input unit, a voice input unit, a display unit, a voice output unit, and a character voice recognition / output control unit, wherein the character voice recognition / output control unit is operated by the character input unit. A character recognition unit that recognizes an input character and outputs character data, a rule voice synthesis unit that converts the character data of the character recognition unit into voice data and outputs the voice data to the voice output unit, and the voice input. A voice recognition unit for recognizing the voice input by the means and outputting voice data; a display unit for converting the voice data of the voice recognition unit into character data and outputting the character data to the display unit; and controlling the respective units. A voice writing device characterized by comprising a control unit for 2. The character recognition unit includes an interface unit,
A feature extraction unit that receives the character data from the character input unit and extracts the feature of the character; a character standard pattern memory;
A collating unit that collates the character data by the feature extraction unit with the character standard pattern of the character standard pattern memory, and determines whether the character data by the feature extraction unit matches the character standard pattern and outputs the matched text data. The voice recognition unit comprises an interface unit, an A / D converter for receiving analog voice data from the voice input unit, and a voice feature of the digital voice data converted by the A / D converter. A voice feature extraction unit that extracts the voice standard pattern memory, a matching unit that matches the voice data of the voice feature extraction unit with the voice standard pattern of the voice standard pattern memory, and the voice data of the voice feature extraction unit. And a determination processing unit that determines the match with the voice standard pattern and outputs the matched text data, The display unit includes an interface unit and a controller that outputs character data to the display unit based on the text data output from the character / speech recognition unit. The rule voice synthesis unit includes an interface unit and a dictionary memory. A sound source memory, a language processing / waveform connection processing unit for converting the text data output from the character recognition unit into a voice waveform based on information in the dictionary memory and the sound source memory, and the language processing / waveform connection processing unit 2. The voice writing device according to claim 1, further comprising a D / A converter for converting the voice data of 1. to analog voice data and outputting the analog voice data to the voice output means. 3. The voice writing according to claim 1, wherein the regular voice synthesizing unit is provided with a recording / playback unit for recording and playing back desired voice data based on input data. vessel. 4. The character input means is a touch panel type character input tablet provided with a character input pen and a keyboard, the display means is a liquid crystal display panel, and the voice input means is a voice input microphone. ,
The voice output means is a voice output speaker, and the character input tablet has a structure in which a display surface and a character input surface are flush with each other so as to be superimposed on the liquid crystal display panel. The voice writing device according to any one of claims 1 to 3.