CN109708283B

CN109708283B - Offline-online mixed voice recognition air conditioning system and implementation method

Info

Publication number: CN109708283B
Application number: CN201910064744.2A
Authority: CN
Inventors: 李昱兵; 唐食明; 郭海峰; 陈立; 张玉蓉; 郭晓东; 梁家琦
Original assignee: Sichuan Hongmei Intelligent Technology Co Ltd
Current assignee: Sichuan Hongmei Intelligent Technology Co Ltd
Priority date: 2019-01-23
Filing date: 2019-01-23
Publication date: 2020-08-04
Anticipated expiration: 2039-01-23
Also published as: CN109708283A

Abstract

The invention provides an off-line-on-line mixed voice recognition air-conditioning system and a realization method, wherein the system receives a recognition name and an equipment number of a voice recognition air-conditioner through a processing server platform, stores the corresponding relation between the recognition name and the equipment number, and analyzes when receiving a voice serial number and a voice instruction to determine a target recognition name and a target equipment number; when receiving a voice instruction input from the outside, the voice recognition air conditioner generates a corresponding voice serial number, and when the voice instruction is self-regulation, the voice recognition air conditioner performs self-regulation and sends the voice serial number and the voice instruction to the processing server platform; when the analyzed voice command and the voice serial number are received, judging whether the execution is executed or not, and if the execution is executed, ignoring; otherwise, executing the analyzed voice command. The scheme provided by the invention can control other voice-controlled air conditioners through one voice-controlled air conditioner, and can further improve the accuracy of the voice command execution of the voice-controlled air conditioner.

Description

Offline-online mixed voice recognition air conditioning system and implementation method

Technical Field

The invention relates to the technical field of air conditioners, in particular to an off-line and on-line mixed voice recognition air conditioner system and an implementation method.

Background

Along with the increasing intellectualization and complication of the air conditioner, the control mode of the remote controller and the control mode of the keys of the machine are inconvenient in the actual use process, and particularly for the old and children, the control mode of the remote controller and the control mode of the keys of the machine are difficult to master. Therefore, the voice-controlled air conditioner with the voice recognition function is produced.

At present, the voice control air conditioner can only carry out single machine control, namely, one voice control air conditioner can only control the air conditioner by itself but not other air conditioners in a voice control mode.

Disclosure of Invention

The embodiment of the invention provides an off-line-on mixed voice recognition air-conditioning system and an implementation method thereof, which can control other voice-controlled air conditioners through one voice-controlled air conditioner and can further improve the accuracy of executing voice commands by the voice-controlled air conditioners.

An offline-online hybrid voice recognition air conditioning system comprising: a processing server platform and at least one voice recognition air conditioner, wherein,

the processing server platform is used for receiving an identification name of the voice recognition air conditioner input from the outside, determining an equipment number corresponding to the voice recognition air conditioner, constructing and storing a corresponding relation between the identification name and the equipment number, and sending the corresponding relation to the voice recognition air conditioner;

the voice recognition air conditioner is used for storing the corresponding relation, generating a corresponding voice serial number for the voice command when receiving an externally input voice command, performing self-regulation and control when determining that the voice command is a regulation and control self command, and sending the voice serial number and the voice command to the processing server platform; when the voice command is determined not to be a command for regulating and controlling the voice command, the voice serial number and the voice command are directly sent to the processing server platform;

the processing server platform is further used for analyzing the voice command, determining a target identification name, determining a target equipment number corresponding to the target identification name according to the corresponding relation and sending the analyzed voice command and the voice serial number to a voice identification air conditioner corresponding to the target equipment number when receiving the voice serial number and the voice command;

the voice recognition air conditioner is further used for judging whether the voice instruction related to the voice serial number is executed or not when the analyzed voice instruction and the voice serial number sent by the processing server platform are received, and if so, ignoring the analyzed voice instruction; otherwise, executing the analyzed voice command.

Preferably, the voice recognition air conditioner includes: a voice conversion unit and a serial number generation unit, wherein,

the voice conversion unit is used for converting the voice instruction into a character, converting the regulation and control instruction into a corresponding regulation and control signal when the character is recognized as the regulation and control instruction for regulating and controlling the voice recognition air conditioner where the character is located, regulating and controlling the voice recognition air conditioner where the character is located through the regulation and control signal, and identifying whether the character contains a target recognition name or not;

the serial number generating unit is used for determining the equipment number of the voice recognition air conditioner where the serial number generating unit is located and the sequence number of the voice command; when the voice conversion unit identifies that the characters contain target identification names, searching for target equipment numbers corresponding to the target identification names according to the corresponding relation, and sequentially arranging preset sending identifiers, equipment numbers of the voice identification air conditioners where the target equipment numbers are located, the target identifiers, the target equipment numbers, the voice identifiers and the sequence numbers of the voice instructions; when the voice conversion unit does not recognize that the characters contain the target recognition name, sequentially arranging a preset sending identifier, the equipment number of the voice recognition air conditioner where the voice conversion unit is located, the target identifier, 0, the voice identifier and the sequence number of the voice command.

Preferably, the first and second electrodes are formed of a metal,

the voice conversion unit is further configured to determine that the text is a regulation and control instruction for regulating and controlling the voice recognition air conditioner where the text is located if the recognized text does not include the target recognition name or the recognized text includes the target recognition name which is the recognition name of the voice recognition air conditioner where the text is located.

Preferably, the processing server platform comprises: an allocation unit, a semantic understanding unit, a voice analysis unit and a serial number correction unit, wherein,

the distribution unit is used for distributing the voice instruction to the semantic understanding unit when the length of the voice instruction is larger than a preset length threshold value, otherwise, distributing the voice instruction to the voice analysis unit;

the semantic understanding unit is used for splitting the voice instruction based on a deep learning algorithm, determining at least one regulating instruction arranged according to a time sequence from a split result, and identifying whether the split result contains a target identification name;

the voice analysis unit is used for converting the voice instruction into characters, identifying a target identification name and a regulation and control instruction contained in the characters, and identifying whether the characters contain the target identification name or not;

the serial number correction unit is used for determining a target equipment number corresponding to the target identification name according to the corresponding relation when the semantic understanding unit/the voice analysis unit identifies the target identification name; judging whether the target equipment number corresponding to the target identification name is consistent with a field between the target identifier and the voice identifier in the voice serial number, if so, keeping the voice serial number unchanged; otherwise, comparing the target equipment number corresponding to the target identification name with a field between the sending identifier and the target identifier in the voice serial number, and keeping the voice serial number unchanged when the comparison result is consistent and the field between the target identifier and the voice identifier is 0; and when the comparison result is inconsistent, replacing a field between the target identifier and the voice identifier in the voice serial number by using the target equipment number corresponding to the target identification name.

Preferably, the first and second electrodes are formed of a metal,

the voice recognition air conditioner is further configured to start timing after the self-regulation, and cancel the self-regulation if the analyzed voice command which is consistent with the voice sequence number corresponding to the self-regulation is not received when the timing duration reaches a preset timing threshold.

Preferably, the voice recognition air conditioner further comprises: a text conversion unit, a power amplifier, and a speaker, wherein,

the character conversion unit is used for converting the regulating instruction into a voice signal;

the power amplifier is used for amplifying the voice signal converted by the character conversion unit;

and the loudspeaker is used for playing the voice signal amplified by the power amplifier.

An implementation method of off-line-on-line hybrid speech recognition comprises the following steps:

receiving an identification name of an externally input voice identification air conditioner through a processing server platform, determining an equipment number corresponding to the voice identification air conditioner, constructing and storing a corresponding relation between the identification name and the equipment number, and sending the corresponding relation to the voice identification air conditioner;

storing the corresponding relation through the voice recognition air conditioner, generating a corresponding voice serial number for a voice instruction when the voice instruction input from the outside is received, performing self-regulation and control when the voice instruction is determined to be a regulation and control self instruction, and sending the voice serial number and the voice instruction to the processing server platform; when the voice command is determined not to be a command for regulating and controlling the voice command, the voice serial number and the voice command are directly sent to the processing server platform;

when the processing server platform receives the voice serial number and the voice command, the voice command is analyzed, a target identification name is determined, a target equipment number corresponding to the target identification name is determined according to the corresponding relation, and the analyzed voice command and the voice serial number are sent to a voice identification air conditioner corresponding to the target equipment number;

when the voice recognition air conditioner receives the analyzed voice command and the voice serial number sent by the processing server platform, judging whether the voice command related to the voice serial number is executed or not, and if so, ignoring the analyzed voice command; otherwise, executing the analyzed voice command.

Preferably, after the voice recognition air conditioner receives an externally input voice command, before generating a corresponding voice serial number for the voice command, the method further includes:

the voice recognition air conditioner converts the voice instruction into a character, when the character is recognized as a regulation and control instruction for regulating and controlling the voice recognition air conditioner where the character is located, the regulation and control instruction is converted into a corresponding regulation and control signal, the voice recognition air conditioner where the character is located is regulated and controlled through the regulation and control signal, and whether the character contains a target recognition name or not is recognized;

the generating of the corresponding voice sequence number for the voice instruction includes:

determining the equipment number of the self voice recognition air conditioner and the sequence number of the voice command;

when the characters contain target identification names, searching target equipment numbers corresponding to the target identification names according to the corresponding relation, and sequentially arranging preset sending identifiers, equipment numbers of the voice identification air conditioners where the target equipment numbers are located, the target identifiers, the target equipment numbers, the voice identifiers and sequence numbers of the voice instructions;

and when the target identification name is not identified in the characters, sequentially arranging a preset sending identifier, the equipment number of the voice identification air conditioner where the character is located, the target identifier, 0, the voice identifier and the sequence number of the voice command.

Preferably, the first and second electrodes are formed of a metal,

the control instruction for identifying the characters to control the self-located voice recognition air conditioner comprises the following steps:

and if the recognized characters do not contain the target recognition name or the recognized target recognition name contained in the characters is the recognition name of the voice recognition air conditioner where the characters are located, determining the characters as a regulation and control instruction for regulating and controlling the voice recognition air conditioner where the characters are located.

Preferably, the first and second electrodes are formed of a metal,

the processing server platform analyzes the voice command and determines a target identification name, and the steps comprise:

when the length of the voice instruction is larger than a preset length threshold value, the voice instruction is split based on a deep learning algorithm, at least one regulating instruction arranged according to a time sequence is determined from a split result, and whether a target identification name is included in the split result is identified; when the length of the voice instruction is not larger than a preset length threshold value, converting the voice instruction into characters, identifying a target identification name and a regulation and control instruction contained in the characters, and identifying whether the characters contain the target identification name or not;

after the analyzing the voice command and determining the target identification name, before sending the analyzed voice command and the voice serial number to the voice identification air conditioner corresponding to the target equipment number, the method further includes:

when the target identification name is identified, judging whether a target equipment number corresponding to the target identification name is consistent with a field between the target identifier and the voice identifier in the voice serial number, and if so, keeping the voice serial number unchanged; otherwise, comparing the target equipment number corresponding to the target identification name with a field between the sending identifier and the target identifier in the voice serial number, and keeping the voice serial number unchanged when the comparison result is consistent and the field between the target identifier and the voice identifier is 0; and when the comparison result is inconsistent, replacing a field between the target identifier and the voice identifier in the voice serial number by using the target equipment number corresponding to the target identification name.

Preferably, the first and second electrodes are formed of a metal,

the above method further comprises: and the voice recognition air conditioner converts the regulation and control instruction into a voice signal, amplifies the voice signal and plays the voice signal amplified by the power amplifier.

Preferably, after the self-regulation, further comprising:

and the voice recognition air conditioner starts timing, and when the timing duration reaches a preset timing threshold value, the analyzed voice command which is consistent with the voice serial number corresponding to the self regulation is not received, and the self regulation is cancelled.

The embodiment of the invention provides an off-line-on mixed voice recognition air-conditioning system and an implementation method thereof, wherein the off-line-on mixed voice recognition air-conditioning system comprises: the system comprises a processing server platform and at least one voice recognition air conditioner, wherein the processing server platform is used for receiving an identification name of the voice recognition air conditioner input from the outside, determining an equipment number corresponding to the voice recognition air conditioner, constructing and storing a corresponding relation between the identification name and the equipment number, and sending the corresponding relation to the voice recognition air conditioner; the voice recognition air conditioner is used for storing the corresponding relation, generating a corresponding voice serial number for the voice command when receiving the voice command input from the outside, carrying out self regulation and control when determining that the voice command is a regulation and control self command, and sending the voice serial number and the voice command to the processing server platform; when the voice command is determined not to be the command for regulating and controlling the voice command, the voice serial number and the voice command are directly sent to a processing server platform; the processing server platform is further used for analyzing the voice command, determining a target identification name, determining a target equipment number corresponding to the target identification name according to the corresponding relation and sending the analyzed voice command and the voice serial number to the voice identification air conditioner corresponding to the target equipment number when the voice serial number and the voice command are received; the voice recognition air conditioner is further used for judging whether the voice instruction related to the voice serial number is executed or not when the analyzed voice instruction and the voice serial number sent by the processing server platform are received, and if the voice instruction related to the voice serial number is executed, the analyzed voice instruction is ignored; otherwise, the analyzed voice command is executed, because the processing server platform can establish connection with at least one voice recognition air conditioner, the process of analyzing the voice command sent by the voice air conditioner is an independent process and is not interfered by the voice air conditioner sending the voice command, the analyzed voice command and the voice serial number are sent to a voice recognition air conditioner corresponding to the target equipment number, if the voice recognition air conditioner corresponding to the target equipment number is consistent with the voice air conditioner sending the voice command, namely the voice command related to the voice serial number is executed, the voice command is not executed repeatedly, the voice control air conditioner can control other voice control air conditioners, and can avoid the voice air conditioner from missing the voice command to be executed due to the analysis problem as much as possible, so that the accuracy of the voice command executed by the voice control air conditioner can be further improved.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.

FIG. 1 is a schematic diagram of an off-line and on-line hybrid speech recognition air conditioning system according to an embodiment of the present invention;

FIG. 2 is a schematic diagram of an off-line and on-line hybrid speech recognition air conditioning system according to another embodiment of the present invention;

FIG. 3 is a schematic diagram of an off-line and on-line hybrid speech recognition air conditioning system according to another embodiment of the present invention;

FIG. 4 is a schematic diagram of an off-line and on-line hybrid speech recognition air conditioning system according to another embodiment of the present invention;

FIG. 5 is a flow chart of a method for implementing off-line-on hybrid speech recognition according to an embodiment of the present invention;

fig. 6 is a flowchart of an implementation method of offline-online hybrid speech recognition according to another embodiment of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer and more complete, the technical solutions in the embodiments of the present invention will be described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention, and based on the embodiments of the present invention, all other embodiments obtained by a person of ordinary skill in the art without creative efforts belong to the scope of the present invention.

As shown in fig. 1, an embodiment of the present invention provides an offline-online hybrid voice recognition air-conditioning system, which may include: a processing server platform 101 and at least one voice recognition air conditioner 102, wherein,

the processing server platform 101 is configured to receive an identification name of the voice recognition air conditioner 102 input from the outside, determine an equipment number corresponding to the voice recognition air conditioner 102, construct and store a correspondence between the identification name and the equipment number, and send the correspondence to the voice recognition air conditioner 102;

the voice recognition air conditioner 102 is configured to store the correspondence, generate a corresponding voice serial number for the voice instruction when receiving the externally input voice instruction, perform self-regulation when determining that the voice instruction is a regulation self instruction, and send the voice serial number and the voice instruction to the processing server platform 101; when the voice command is determined not to be a command for regulating and controlling itself, the voice serial number and the voice command are directly sent to the processing server platform 101;

the processing server platform 101 is further configured to, when receiving the voice serial number and the voice command, parse the voice command, determine a target identification name, determine a target device number corresponding to the target identification name according to the correspondence, and send the parsed voice command and the voice serial number to the voice recognition air conditioner 102 corresponding to the target device number;

the voice recognition air conditioner 102 is further configured to, when receiving the analyzed voice command and the voice serial number sent by the processing server platform 101, determine whether the voice command related to the voice serial number has been executed, and if so, ignore the analyzed voice command; otherwise, executing the analyzed voice command.

The recognition name of the voice recognition air conditioner received by the processing server platform from the outside can be the recognition name of the voice recognition air conditioner received by the client, and the recognition name can be a main lying air conditioner, a secondary lying 1 air conditioner, a living room air conditioner and the like.

In addition, the device number corresponding to the voice recognition air conditioner is a feature identifier of the voice recognition air conditioner, each voice recognition air conditioner has its own device number when leaving the factory, the process of determining the voice recognition air conditioner may be input by a user through a client, or after the voice recognition air conditioner is bound to the client, the voice recognition air conditioner is sent to the client, and then the client sends the device number corresponding to the voice recognition air conditioner to the processing server platform.

In addition, the above-mentioned corresponding relation can be constructed by a table, namely, two items in the table are named as an identification name and an equipment number respectively, and the corresponding relation is constructed for the two items in the table.

In addition, the corresponding relation stored by the voice recognition air conditioner can be the self recognition name and the self equipment number; or the identification names and equipment numbers corresponding to all the voice recognition air conditioners bound to the same registered account number with the voice recognition air conditioners.

The voice sequence number is a characteristic identifier of the voice command, and one voice command corresponds to one voice sequence number. The processing server platform determines the number of the target equipment according to the target identification name contained in the analyzed voice command, and sends the analyzed voice command to the equipment corresponding to the target equipment number, and whether the target equipment number is the voice recognition air conditioner for sending the voice command or not can not be distinguished, because the voice recognition air conditioner possibly has analysis errors on the voice command, the command aiming at the voice recognition air conditioner can not be analyzed, and the probability of wrong execution of the voice command can be greatly reduced by sending the voice command again through the processing server platform. The voice recognition air conditioner judges whether a voice instruction related to the voice serial number is executed or not, and if so, ignores the analyzed voice instruction; otherwise, the parsed voice command is executed, and the process can avoid the voice command from being repeatedly executed.

In the embodiment shown in fig. 1, the offline-online hybrid voice recognition air conditioning system includes: the system comprises a processing server platform and at least one voice recognition air conditioner, wherein the processing server platform is used for receiving an identification name of the voice recognition air conditioner input from the outside, determining an equipment number corresponding to the voice recognition air conditioner, constructing and storing a corresponding relation between the identification name and the equipment number, and sending the corresponding relation to the voice recognition air conditioner; the voice recognition air conditioner is used for storing the corresponding relation, generating a corresponding voice serial number for the voice command when receiving the voice command input from the outside, carrying out self regulation and control when determining that the voice command is a regulation and control self command, and sending the voice serial number and the voice command to the processing server platform; when the voice command is determined not to be the command for regulating and controlling the voice command, the voice serial number and the voice command are directly sent to a processing server platform; the processing server platform is further used for analyzing the voice command, determining a target identification name, determining a target equipment number corresponding to the target identification name according to the corresponding relation and sending the analyzed voice command and the voice serial number to the voice identification air conditioner corresponding to the target equipment number when the voice serial number and the voice command are received; the voice recognition air conditioner is further used for judging whether the voice instruction related to the voice serial number is executed or not when the analyzed voice instruction and the voice serial number sent by the processing server platform are received, and if the voice instruction related to the voice serial number is executed, the analyzed voice instruction is ignored; otherwise, the analyzed voice command is executed, because the processing server platform can establish connection with at least one voice recognition air conditioner, the process of analyzing the voice command sent by the voice air conditioner is an independent process and is not interfered by the voice air conditioner sending the voice command, the analyzed voice command and the voice serial number are sent to a voice recognition air conditioner corresponding to the target equipment number, if the voice recognition air conditioner corresponding to the target equipment number is consistent with the voice air conditioner sending the voice command, namely the voice command related to the voice serial number is executed, the voice command is not executed repeatedly, the voice control air conditioner can control other voice control air conditioners, and can avoid the voice air conditioner from missing the voice command to be executed due to the analysis problem as much as possible, so that the accuracy of the voice command executed by the voice control air conditioner can be further improved.

In another embodiment of the present invention, as shown in fig. 2, the off-line-on-line hybrid speech recognition air conditioning system, the speech recognition air conditioner 102, includes: a voice conversion unit 1021, and a serial number generation unit 1022, wherein,

the voice conversion unit 1021 is used for converting the voice instruction into a character, converting the regulation instruction into a corresponding regulation signal when the character is recognized as the regulation instruction for regulating the voice recognition air conditioner where the character is located, regulating the voice recognition air conditioner where the character is located through the regulation signal, and recognizing whether the character contains a target recognition name;

a serial number generating unit 1022, configured to determine a device number of the speech recognition air conditioner where the serial number is located and a sequence number of the speech instruction; when the voice conversion unit 1021 recognizes that the characters contain the target recognition name, searching for a target equipment number corresponding to the target recognition name according to the corresponding relation, and sequentially arranging a preset sending identifier, an equipment number of the voice recognition air conditioner where the voice recognition air conditioner is located, the target identifier, the target equipment number, the voice identifier and a sequence number of the voice command; when the voice conversion unit 1021 does not recognize that the text contains the target recognition name, the preset sending identifier, the equipment number of the voice recognition air conditioner where the voice conversion unit is located, the target identifier, 0, the voice identifier and the sequence number of the voice command are sequentially arranged.

The control instruction for controlling the voice recognition air conditioner in which the character is recognized can be that the recognized character contains the self-recognition name or the recognized character does not contain any recognition name.

In addition, when the characters are recognized to contain target recognition names, the voice serial numbers contain preset sending identifiers, the equipment numbers of the voice recognition air conditioners where the voice serial numbers are located, the target identifiers, the target equipment numbers, the voice identifiers and the sequence numbers of the voice instructions; when the characters do not comprise the target identification names, the preset sending identifiers, the equipment numbers of the voice identification air conditioners where the characters are located, the target identifiers, 0, the voice identifiers and the sequence numbers of the voice instructions are sequentially arranged, and the uniqueness of the voice sequence numbers is guaranteed. Meanwhile, the voice recognition air conditioner which sends the voice command to the processing server platform can be found out from the voice serial number, and the target voice recognition air conditioner which sends the voice command to the processing server platform can also be found out.

In another embodiment of the present invention, the voice converting unit is further configured to determine that the text is a control instruction for controlling the voice recognition air conditioner where the text is located if the recognized text does not include the target recognition name or the target recognition name included in the recognized text is the recognition name of the voice recognition air conditioner where the text is located.

In another embodiment of the present invention, as shown in fig. 3, the processing server platform 101 includes: an allocation unit 1011, a semantic understanding unit 1012, a speech parsing unit 1013, and a sequence number modification unit 1014, wherein,

an allocating unit 1011, configured to allocate the voice instruction to the semantic understanding unit 1012 when the length of the voice instruction is greater than a preset length threshold, and otherwise, allocate the voice instruction to the voice parsing unit 1013;

a semantic understanding unit 1012, configured to split the voice instruction based on a deep learning algorithm, determine, from a split result, at least one control instruction arranged according to a time sequence, and identify whether the split result includes a target identification name;

the voice analyzing unit 1013 is configured to convert the voice instruction into a text, identify a target identification name and a control instruction included in the text, and identify whether the text includes the target identification name;

a serial number correction unit 1014 configured to determine, when the semantic understanding unit 1012/the voice parsing unit 1013 recognizes the target identification name, a target device number corresponding to the target identification name according to the correspondence; judging whether the target equipment number corresponding to the target identification name is consistent with a field between the target identifier and the voice identifier in the voice serial number or not, and if so, keeping the voice serial number unchanged; otherwise, comparing the target equipment number corresponding to the target identification name with a field between the sending identifier and the target identifier in the voice serial number, and keeping the voice serial number unchanged when the comparison result is consistent and the field between the target identifier and the voice identifier is 0; and when the comparison result is inconsistent, replacing the field between the target identifier and the voice identifier in the voice serial number by using the target equipment number corresponding to the target identification name.

The semantic understanding unit is used for performing voice instruction analysis based on a deep learning method, and the voice analysis unit is used for determining a bid identification name and a regulation instruction by comparing the voice instruction with a pre-stored instruction and a pre-stored identification name based on pre-stored single regulation instructions such as 'raise once', 'lower once' and the like and the identification name. The deep learning method can identify more complicated sentences such as "raise the first degree humidity to adjust to 40% wind speed to be three-level wind direction to be up and down" and the like. The voice command is analyzed through the two modes, so that the analysis accuracy of long sentences, namely the length of the voice command is greater than a preset length threshold value is guaranteed while the analysis rate of the simple voice command is guaranteed.

In another embodiment of the present invention, the voice recognition air conditioner 102 is further configured to start timing after self-regulation, and cancel self-regulation if the timing duration reaches a preset timing threshold value and an analyzed voice command that is consistent with a voice serial number corresponding to self-regulation is not received. By the process, the voice command analyzed by the voice recognition air conditioner is corrected, so that the accuracy of executing the voice command is ensured.

The cancellation of the self-regulation refers to cancellation of the regulation according to the voice instruction, for example, if the voice instruction is "once increased", after the voice instruction is executed, the cancellation of the self-regulation is to perform the regulation once decreased, so that the state of the voice-regulated air conditioner is maintained in the state before the voice instruction is received.

In another embodiment of the present invention, as shown in fig. 4, the voice recognition air conditioner further includes: a text conversion unit 1023, a power amplifier 1024, and a speaker 1025, wherein,

the character conversion unit 1023 is used for converting the regulation and control instruction into a voice signal;

a power amplifier 1024 for amplifying the voice signal converted by the text conversion unit 1023;

and the loudspeaker 1025 is used for playing the voice signal amplified by the power amplifier 1024.

Through the process, the execution result is played, and a user can conveniently learn whether the instruction under the user is executed.

As shown in fig. 5, an embodiment of the present invention provides an implementation method of offline-online hybrid speech recognition, which may include the following steps:

step 501: receiving an identification name of an externally input voice identification air conditioner through a processing server platform, determining an equipment number corresponding to the voice identification air conditioner, constructing and storing a corresponding relation between the identification name and the equipment number, and sending the corresponding relation to the voice identification air conditioner;

step 502: storing the corresponding relation through a voice recognition air conditioner, generating a corresponding voice serial number for a voice instruction when the voice instruction input from the outside is received, performing self-regulation and control when the voice instruction is determined to be a regulation and control self instruction, and sending the voice serial number and the voice instruction to a processing server platform; when the voice command is determined not to be the command for regulating and controlling the voice command, the voice serial number and the voice command are directly sent to a processing server platform;

step 503: when receiving the voice serial number and the voice command, the processing server platform analyzes the voice command, determines a target identification name, determines a target equipment number corresponding to the target identification name according to the corresponding relation, and sends the analyzed voice command and the voice serial number to a voice identification air conditioner corresponding to the target equipment number;

step 504: when the voice recognition air conditioner receives the analyzed voice command and the voice serial number sent by the processing server platform, judging whether the voice command related to the voice serial number is executed or not, and if so, ignoring the analyzed voice command; otherwise, executing the analyzed voice command.

In another embodiment of the present invention, after the voice recognition air conditioner receives the voice command inputted from the outside, before the generating the corresponding voice serial number for the voice command, the method further includes: the voice recognition air conditioner converts the voice instruction into a character, when the character is recognized as a regulation and control instruction for regulating and controlling the voice recognition air conditioner where the character is located, the regulation and control instruction is converted into a corresponding regulation and control signal, the voice recognition air conditioner where the character is located is regulated and controlled through the regulation and control signal, and whether the character contains a target recognition name or not is recognized; the generating of the corresponding voice sequence number for the voice command includes: determining the equipment number of the self voice recognition air conditioner and the sequence number of the voice command; when the characters contain target identification names, searching target equipment numbers corresponding to the target identification names according to the corresponding relation, and sequentially arranging preset sending identifiers, equipment numbers of the voice identification air conditioners where the equipment numbers are located, the target identifiers, the target equipment numbers, the voice identifiers and sequence numbers of voice instructions; and when the character is not recognized to contain the target recognition name, sequentially arranging a preset sending identifier, the equipment number of the voice recognition air conditioner where the character is located, the target identifier, 0, the voice identifier and the sequence number of the voice command.

In an embodiment of the present invention, the recognizing the text as a control instruction for controlling the voice recognition air conditioner in which the text is located includes: and if the recognized characters do not contain the target recognition name or the target recognition name contained in the recognized characters is the recognition name of the voice recognition air conditioner where the characters are located, determining the characters as a regulation and control instruction for regulating and controlling the voice recognition air conditioner where the characters are located.

In an embodiment of the present invention, the parsing the voice command and determining the target identification name by the processing server platform includes: when the length of the voice instruction is larger than a preset length threshold value, the voice instruction is split based on a deep learning algorithm, at least one regulating instruction arranged according to a time sequence is determined from a split result, and whether a target identification name is included in the split result is identified; when the length of the voice instruction is not greater than a preset length threshold value, converting the voice instruction into characters, identifying a target identification name and a regulation and control instruction contained in the characters, and identifying whether the characters contain the target identification name or not; after the voice command is analyzed and the target identification name is determined, before the analyzed voice command and the voice serial number are sent to the voice identification air conditioner corresponding to the target equipment number, the method further comprises the following steps: when the target identification name is identified, judging whether the target equipment number corresponding to the target identification name is consistent with a field between the target identifier and the voice identifier in the voice serial number, and if so, keeping the voice serial number unchanged; otherwise, comparing the target equipment number corresponding to the target identification name with a field between the sending identifier and the target identifier in the voice serial number, and keeping the voice serial number unchanged when the comparison result is consistent and the field between the target identifier and the voice identifier is 0; and when the comparison result is inconsistent, replacing the field between the target identifier and the voice identifier in the voice serial number by using the target equipment number corresponding to the target identification name.

In an embodiment of the present invention, the implementation method further includes: the voice recognition air conditioner converts the regulation and control instruction into a voice signal, amplifies the voice signal and plays the voice signal amplified by the power amplifier.

In one embodiment of the present invention, after self-regulation, further comprising: and starting timing by the voice recognition air conditioner, and when the timing duration reaches a preset timing threshold value, not receiving the analyzed voice command which is consistent with the voice serial number corresponding to the self regulation and control, and then canceling the self regulation and control.

In order to clearly illustrate the above method for implementing offline-online hybrid voice recognition, the following description will be made by taking the communication between the voice recognition air conditioner 1 and the voice recognition air conditioner 2 bound to the same account and the processing server platform as an example. As shown in fig. 6, the method for implementing offline-online hybrid speech recognition may include the following steps:

step 601: receiving an identification name of an externally input voice identification air conditioner through a processing server platform, and determining an equipment number corresponding to the voice identification air conditioner;

for example, the identification names of the voice recognition air conditioner 1 and the voice recognition air conditioner 2 bound to the same account are sent to the processing server platform, and may be sent through the registered client. The recognition name of the voice recognition air conditioner 1 is a living room air conditioner, and the recognition name of the voice recognition air conditioner 2 is a main lying air conditioner; the equipment number corresponding to the voice recognition air conditioner is determined, the equipment number can be sent through the client side, and the equipment number of the voice recognition air conditioner can be obtained through scanning the two-dimensional code arranged on the voice recognition air conditioner through the client side.

Step 602: constructing and storing a corresponding relation between the identification name and the equipment number, and sending the corresponding relation to the voice identification air conditioner;

the correspondence may be constructed in a tabular manner. Sending the corresponding relationship to the voice recognition air conditioner means sending the corresponding relationship corresponding to the same account to all the voice recognition air conditioners bound to the same account, for example, sending both the corresponding relationship of the voice recognition air conditioner 1 and the corresponding relationship of the voice recognition air conditioner 2 bound to the same account to the voice recognition air conditioner 1 and the voice recognition air conditioner 2.

Step 603: when receiving an externally input voice instruction, the voice recognition air conditioner converts the voice instruction into characters and generates a corresponding voice serial number for the voice instruction; when the character is recognized as a regulation and control instruction for regulating and controlling the voice recognition air conditioner, executing step 604; when the voice command is determined not to be the command of the control body, step 607 is executed;

the specific implementation process of generating the corresponding voice sequence number for the voice command is as follows: identifying whether the characters contain target identification names or not; determining the equipment number of the self voice recognition air conditioner and the sequence number of the voice command; when the characters contain target identification names, searching target equipment numbers corresponding to the target identification names according to the corresponding relation, and sequentially arranging preset sending identifiers, equipment numbers of the voice identification air conditioners where the target equipment numbers are located, the target identifiers, the target equipment numbers, the voice identifiers and sequence numbers of voice instructions; and when the character is not recognized to contain the target recognition name, sequentially arranging a preset sending identifier, the equipment number of the voice recognition air conditioner where the character is located, the target identifier, 0, the voice identifier and the sequence number of the voice command.

If the recognized characters do not contain the target recognition name or the recognized characters contain the target recognition name which is the recognition name of the voice recognition air conditioner where the characters are located, determining the characters as a regulation and control instruction for regulating and controlling the voice recognition air conditioner where the characters are located;

step 604: converting the regulation instruction into a corresponding regulation signal, regulating and controlling the voice recognition air conditioner in which the regulation instruction is positioned through the regulation signal, and respectively executing the step 605 and the step 607;

step 605: the voice recognition air conditioner starts timing, judges whether an analyzed voice command which is consistent with a voice serial number corresponding to self regulation and control is received or not when the timing duration reaches a preset timing threshold, and executes a step 615 if the analyzed voice command is consistent with the voice serial number corresponding to self regulation and control; otherwise, go to step 606;

the process is to avoid the recognition error of the voice recognition air conditioner, further analyze the voice recognition air conditioner through the processing server platform, and ensure the accuracy of the voice recognition. Such as: the voice recognition air conditioner 1 receives a voice instruction, the voice recognition air conditioner 1 recognizes that the self regulation and control is performed, however, the timing is started, the timing duration reaches the preset timing threshold, the voice recognition air conditioner 1 still does not receive the analyzed voice instruction which is consistent with the voice serial number corresponding to the self regulation and control again, the recognition error of the voice recognition air conditioner 1 is indicated, and the self regulation and control is cancelled.

Step 606: canceling self regulation and control, and ending the current flow;

for example, if the self-regulation is increased once and the self-regulation is cancelled, the temperature is decreased once to restore to the state before the self-regulation.

Step 607: the voice recognition air conditioner sends the voice serial number and the voice command to a processing server platform;

step 608: the processing server platform receives the voice serial number and the voice command, and executes step 609 when the length of the voice command is greater than a preset length threshold; when the length of the voice command is not greater than the preset length threshold, executing step 610;

the length threshold may be set by the user himself.

Step 609: splitting the voice instruction based on a deep learning algorithm, determining at least one regulating instruction arranged according to a time sequence from a splitting result, identifying whether the splitting result contains a target identification name, and if so, executing step 611; otherwise, go to step 612;

and a deep learning algorithm is adopted for a relatively long sentence, so that the identification accuracy is ensured. Such as: one sentence includes that the air conditioner of the main-lying 1 is raised for one degree, then the humidity is adjusted to 40%, and the wind direction is adjusted to the left and right, and the splitting and the combining analysis are carried out based on the deep learning algorithm due to the longer length. The deep learning algorithm is the prior art and is not described herein.

Whether the target identification name is included in the result of the identification and splitting is mainly used for correcting the voice sequence number.

Step 610: converting the voice instruction into a text, identifying a target identification name and a regulation and control instruction contained in the text, identifying whether the text contains the target identification name, and if so, executing step 611; otherwise, go to step 612;

step 611: correcting the voice sequence number according to the target identification name, and executing step 613;

the specific implementation manner of the step is as follows: judging whether the target equipment number corresponding to the target identification name is consistent with a field between the target identifier and the voice identifier in the voice serial number or not, and if so, keeping the voice serial number unchanged; otherwise, comparing the target equipment number corresponding to the target identification name with a field between the sending identifier and the target identifier in the voice serial number, and keeping the voice serial number unchanged when the comparison result is consistent and the field between the target identifier and the voice identifier is 0; and when the comparison result is inconsistent, replacing the field between the target identifier and the voice identifier in the voice serial number by using the target equipment number corresponding to the target identification name.

Step 612: keeping the voice sequence number unchanged;

step 613: sending the analyzed regulating and controlling instruction and the corresponding voice serial number to a voice recognition air conditioner;

the sent voice recognition air conditioner is the voice recognition air conditioner of the target equipment number corresponding to the target recognition name recognized in the step, and if the target recognition name is not recognized, the voice recognition air conditioner is sent according to the voice serial number identification.

Step 614: when receiving the analyzed voice command and the voice serial number sent by the processing server platform, the voice recognition air conditioner judges whether the voice command related to the voice serial number is executed, if so, the step 615 is executed; otherwise, go to step 616;

the process of the judgment is to avoid the voice recognition air conditioner from repeatedly executing the voice command.

Step 615: ignoring the parsed voice command and performing step 617;

step 616: executing the analyzed voice command;

step 617: the voice recognition air conditioner converts the regulation and control instruction into a voice signal, amplifies the voice signal and plays the voice signal amplified by the power amplifier.

It can be understood that, the above-mentioned implementation of online voice recognition and the control of different voice recognition air conditioners are completed for the voice recognition air conditioners under the same account, but the voice recognition air conditioners under different accounts cannot be controlled mutually, for example: the air conditioner 1 and the air conditioner 2 both belong to the account A, and then the air conditioner 1 can receive a voice command for controlling the air conditioner 2 and realize controlling the air conditioner 2 through online voice recognition.

Embodiments of the present invention provide a readable medium, which includes an execution instruction, and when a processor of a storage controller executes the execution instruction, the storage controller executes a method provided in any one of the above embodiments of the present invention.

An embodiment of the present invention provides a storage controller, including: a processor, a memory, and a bus; the memory is used for storing execution instructions, the processor is connected with the memory through the bus, and when the storage controller runs, the processor executes the execution instructions stored in the memory, so that the storage controller executes the method provided by any one of the above embodiments of the invention.

In summary, the above embodiments of the present invention have at least the following advantages:

1. in an embodiment of the present invention, an offline-online hybrid speech recognition air conditioning system includes: the system comprises a processing server platform and at least one voice recognition air conditioner, wherein the processing server platform is used for receiving an identification name of the voice recognition air conditioner input from the outside, determining an equipment number corresponding to the voice recognition air conditioner, constructing and storing a corresponding relation between the identification name and the equipment number, and sending the corresponding relation to the voice recognition air conditioner; the voice recognition air conditioner is used for storing the corresponding relation, generating a corresponding voice serial number for the voice command when receiving the voice command input from the outside, carrying out self regulation and control when determining that the voice command is a regulation and control self command, and sending the voice serial number and the voice command to the processing server platform; when the voice command is determined not to be the command for regulating and controlling the voice command, the voice serial number and the voice command are directly sent to a processing server platform; the processing server platform is further used for analyzing the voice command, determining a target identification name, determining a target equipment number corresponding to the target identification name according to the corresponding relation and sending the analyzed voice command and the voice serial number to the voice identification air conditioner corresponding to the target equipment number when the voice serial number and the voice command are received; the voice recognition air conditioner is further used for judging whether the voice instruction related to the voice serial number is executed or not when the analyzed voice instruction and the voice serial number sent by the processing server platform are received, and if the voice instruction related to the voice serial number is executed, the analyzed voice instruction is ignored; otherwise, the analyzed voice command is executed, because the processing server platform can establish connection with at least one voice recognition air conditioner, the process of analyzing the voice command sent by the voice air conditioner is an independent process and is not interfered by the voice air conditioner sending the voice command, the analyzed voice command and the voice serial number are sent to a voice recognition air conditioner corresponding to the target equipment number, if the voice recognition air conditioner corresponding to the target equipment number is consistent with the voice air conditioner sending the voice command, namely the voice command related to the voice serial number is executed, the voice command is not executed repeatedly, the voice control air conditioner can control other voice control air conditioners, and can avoid the voice air conditioner from missing the voice command to be executed due to the analysis problem as much as possible, so that the accuracy of the voice command executed by the voice control air conditioner can be further improved.

2. In the embodiment of the invention, when the characters are recognized to contain the target recognition name, the voice serial number contains a preset sending identifier, the equipment number of the voice recognition air conditioner where the voice serial number is located, the target identifier, the target equipment number, the voice identifier and the sequence number of the voice command; when the characters do not comprise the target identification names, the preset sending identifiers, the equipment numbers of the voice identification air conditioners where the characters are located, the target identifiers, 0, the voice identifiers and the sequence numbers of the voice instructions are sequentially arranged, and the uniqueness of the voice sequence numbers is guaranteed. Meanwhile, the voice recognition air conditioner which sends the voice command to the processing server platform can be found out from the voice serial number, and the target voice recognition air conditioner which sends the voice command can also be found out, so that the voice serial number has multiple purposes.

3. In the embodiment of the invention, in the processing server platform, for the longer voice instruction, the voice instruction is analyzed based on the deep learning method, and for the shorter voice instruction, based on some single pre-stored regulation and control instructions and identification names, the voice instruction is compared with the pre-stored instruction and identification names to determine the identification name for sending the label and the regulation and control instruction. Deep learning methods can identify more complex sentences. The voice command is analyzed through the two modes, so that the analysis accuracy of long sentences, namely the length of the voice command is greater than a preset length threshold value is guaranteed while the analysis rate of the simple voice command is guaranteed.

4. In the embodiment of the present invention, the voice recognition air conditioner is further configured to start timing after self-regulation, and cancel self-regulation if an analyzed voice command that is consistent with a voice sequence number corresponding to self-regulation is not received when a timing duration reaches a preset timing threshold. By the process, the voice command analyzed by the voice recognition air conditioner is corrected, so that the accuracy of executing the voice command is ensured.

It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising a" does not exclude the presence of other similar elements in a process, method, article, or apparatus that comprises the element.

Those of ordinary skill in the art will understand that: all or part of the steps for realizing the method embodiments can be completed by hardware related to program instructions, the program can be stored in a computer readable storage medium, and the program executes the steps comprising the method embodiments when executed; and the aforementioned storage medium includes: various media that can store program codes, such as ROM, RAM, magnetic or optical disks.

Finally, it is to be noted that: the above description is only a preferred embodiment of the present invention, and is only used to illustrate the technical solutions of the present invention, and not to limit the protection scope of the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention shall fall within the protection scope of the present invention.

Claims

1. An off-line-on-line hybrid speech recognition air conditioning system, comprising: a processing server platform and at least one voice recognition air conditioner, wherein,

2. The off-line and on-line hybrid voice recognition air conditioning system of claim 1, wherein the voice recognition air conditioner comprises: a voice conversion unit and a serial number generation unit, wherein,

3. The off-line and on-line hybrid voice recognition air conditioning system of claim 2,

4. The offline-online hybrid voice recognition air conditioning system of claim 2, said processing server platform comprising: an allocation unit, a semantic understanding unit, a voice analysis unit and a serial number correction unit, wherein,

5. The off-line and on-line hybrid voice recognition air conditioning system according to any one of claims 1 to 4,

6. The off-line and on-line hybrid voice recognition air conditioning system according to any one of claims 2 to 4, wherein the voice recognition air conditioner further comprises: a text conversion unit, a power amplifier, and a speaker, wherein,

7. An implementation method of offline-online hybrid speech recognition, comprising:

8. The method for implementing offline-online hybrid speech recognition according to claim 7, wherein after the speech recognition air conditioner receives an externally input speech command, before the generating a corresponding speech sequence number for the speech command, the method further comprises:

9. The method of claim 8,

and/or the presence of a gas in the gas,

when the target identification name is identified, judging whether a target equipment number corresponding to the target identification name is consistent with a field between the target identifier and the voice identifier in the voice serial number, and if so, keeping the voice serial number unchanged; otherwise, comparing the target equipment number corresponding to the target identification name with a field between the sending identifier and the target identifier in the voice serial number, and keeping the voice serial number unchanged when the comparison result is consistent and the field between the target identifier and the voice identifier is 0; when the comparison result is inconsistent, replacing a field between the target identifier and the voice identifier in the voice serial number by using the target equipment number corresponding to the target identification name;

and/or the presence of a gas in the gas,

further comprising: and the voice recognition air conditioner converts the regulation and control instruction into a voice signal, amplifies the voice signal and plays the voice signal amplified by the power amplifier.

10. The method for implementing offline-online hybrid speech recognition according to any one of claims 7 to 9, further comprising, after the self-regulation: