JP2008076812A - Voice recognition device, voice recognition method and voice recognition program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a voice recognition device, a voice recognition method and a voice recognition program, capable of improving recognition accuracy of utterance by suitably grasping a keyword, even when user's free utterance is recognized. <P>SOLUTION: The voice recognition device 1 comprises: a text conversion means 11 for recognizing input voice and converting it to a text; a first syntax analysis means 31 for outputting a result of analysis processing of the text by using a word string score, together with the score, as a first voice candidate group; a second syntax analysis means 32 for outputting the result of analysis processing of the text by using the word string score, together with the score as a second voice candidate group; and a voice candidate group determination means 33 for determining a final voice candidate group, based on the score from the first voice candidate and the second voice candidate. When it is judged by a comparison judgement means 34 that the first voice candidate with the highest score is not coincident with the second candidate with the highest score, the highest score of the second voice candidate is increased by the prescribed value amount. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a speech recognition device that recognizes input speech, a speech recognition method, and a speech recognition program.

  In recent years, for example, in a system in which a user operates a device, a voice recognition device that recognizes a voice input by the user and acquires information necessary for the operation of the device has been used. In such a voice recognition device, a voice (utterance) input by the user is recognized, and the user's next utterance is prompted in response to the user based on the recognized result. Is done. Information necessary for operating the device is acquired from the result of recognizing the dialogue with the user.

  In this voice recognition apparatus, a voice recognition apparatus that accepts not only a pre-registered command but also a user's free utterance that is not limited in terms of speech as an input voice has been proposed. In this speech recognition apparatus, an input speech is converted from an acoustic feature into a word, and analyzed based on the appearance probability (N-gram) of a word string, and recognized as text represented by the word string (dictation) ) Is used. Here, “text” is a significant syntax expressed in a string of words and having a predetermined meaning. However, in such a speech recognition apparatus that accepts a free utterance, it is necessary to recognize an ambiguous input speech, so that the analysis is difficult and the possibility of erroneous recognition increases. For this reason, a technique has been proposed for improving recognition accuracy by re-recognizing a word or word string that may have been misrecognized by another method with respect to a recognition result by dictation (see, for example, Patent Document 1). .

In the speech recognition apparatus of Patent Document 1, a word or a word string having a certainty factor (calculated from the occurrence probability of a word string) having a predetermined threshold value or less is re-recognized from a dictation recognition result. "Target word string". Then, speech recognition using rules is performed on the speech of the re-recognition target word string. A rule defines whether or not every word string appears instead of the appearance probability. In speech recognition using a rule, only a word string defined to appear can be a recognition result. In the speech recognition apparatus of Patent Literature 1, a predetermined condition to be satisfied for the re-recognition target word string is determined from a portion other than the re-recognition target word string of the dictation recognition result, and the speech is set so as to satisfy this condition using a rule. Recognize. Then, the dictation recognition result is replaced with the voice recognition result using the rule.
JP-A-2005-202165

  In the speech recognition apparatus, it is assumed that a word with low certainty is a misrecognized word in speech recognition by dictation. However, in speech recognition by dictation, just because a certain word has a low certainty does not mean that the word itself has been misrecognized. That is, a word with low confidence in the dictation recognition result means that the word is out of context in the text (sentence) recognized by the dictation. At this time, the fact that the word is output as a recognition result despite being out of context is considered to be a word (keyword) that is audibly and clearly pronounced by the user. For this reason, there is a problem that the recognition accuracy cannot be improved by a method of re-recognizing the voice of a word with a low certainty level and replacing it with another word as in the voice recognition apparatus.

  In view of the above circumstances, the present invention provides a speech recognition apparatus, speech recognition method, and speech recognition that can appropriately recognize a keyword and improve the recognition accuracy of the speech even when a user's free speech is recognized. The purpose is to provide a program.

  The speech recognition apparatus according to the present invention recognizes input speech to convert the speech into text expressed by a word string, and words and words included in the text converted by the text conversion means A first syntax for executing processing for analyzing the text using a word sequence score calculated based on a feature of the sequence, and outputting a result of the processing as a first speech candidate group together with a score of each speech candidate An analysis unit and a process of analyzing the text using a word score calculated based on a feature of the word included in the text converted by the text conversion unit, and the result of the process is used as a second speech candidate group From the second syntactic analysis means for outputting together with the score of each voice candidate, the first voice candidate group, and the second voice candidate group, the final voice based on the score of each voice candidate A speech candidate group determining means for determining a complement group, and a comparison determination means for comparing the first speech candidate group and the second speech candidate group to determine whether there is a matching speech candidate; The first speech candidate that has the highest score among the speech candidates included in the first speech candidate group and the second score that has the highest score among speech candidates included in the second speech candidate group by the comparison determination unit. When the voice candidate is determined not to match, the highest score of the second voice candidate is increased by a predetermined value.

  According to the speech recognition apparatus of the present invention, the first syntax analysis unit is calculated based on the characteristics of the words and word strings included in the text obtained by recognizing the input speech. The process of analyzing the text is executed using the word string score. Here, the “score” is a likelihood that a speech candidate (a word or a word string that is a candidate for the recognition result of the input speech) corresponds to the input speech from various viewpoints such as an acoustic viewpoint and a linguistic viewpoint (likelihood). (Degree of confidence). The “word string score” is calculated based on, for example, the appearance probability of the word and the word string, and the word string score is used to limit not only the pre-registered command but also the wording. The user's free speech can be recognized.

  The second syntax analysis means analyzes the text using the word score calculated based on the characteristics of the words included in the text obtained by recognizing the input speech. Execute the process. The “word score” is calculated based on, for example, the appearance probability for each word.

  At this time, the first speech candidate that has the highest score among the speech candidates included in the first speech candidate group and the second score that has the highest score among speech candidates included in the second speech candidate group. When it is determined that the speech candidates do not match, the present invention sets the final speech candidate group to the speech candidate group reflecting the meaning of the keyword and the function of the keyword. The score of the second speech candidate having the highest score in the group is increased by a predetermined value.

  Then, the speech candidate group determining means determines a final speech candidate group from the first speech candidate group and the second speech candidate group based on the score of each speech candidate. At this time, since the score of the second speech candidate having the highest score in the second speech candidate group is increased, the speech candidate including the meaning of the keyword is ranked higher in the final speech candidate group. Increases the chance of entering. Therefore, according to the speech recognition apparatus of the present invention, for example, even if a user's natural utterance is input, the keyword is properly grasped, and thus the recognition accuracy of the utterance can be improved.

  In the speech recognition apparatus of the present invention, the speech candidate group determination means finally selects a speech candidate group whose score is within a predetermined rank among speech candidates included in the first and second speech candidate groups. It is preferable to determine the speech candidate group.

  In this case, since the score of the second speech candidate that is considered to include the meaning of the keyword is increased, the second speech candidate is a speech candidate whose score is within a predetermined rank (for example, the highest score). Voice candidate). Therefore, for example, even if a user's natural utterance is input, the final speech candidate group can be determined by appropriately grasping the keyword, and the recognition accuracy of the utterance can be improved.

  Further, in the speech recognition apparatus of the present invention, a notification unit that notifies a plurality of speech candidates included in the first and second speech candidate groups based on the scores of the respective speech candidates, and a plurality of notifications notified by the notification unit Preferably, the voice candidate group determination unit determines the voice candidate group selected by the selection unit as a final voice candidate group.

  In this case, since the score of the second speech candidate that is considered to include the meaning of the keyword has been increased, the speech candidates corresponding to the user's utterance can be included with a high probability in the plurality of speech candidates to be notified. The user can select a speech candidate according to the intention. Therefore, for example, even if a user's natural utterance is input, the final speech candidate group can be determined by appropriately grasping the keyword, and the recognition accuracy of the utterance can be improved.

  The speech recognition apparatus of the present invention preferably includes a control unit that executes a predetermined control process based on at least the final speech candidate group determined by the speech candidate group determination unit.

  In this case, based on the final speech candidate group, for example, a predetermined control process is determined and executed from a plurality of predetermined control processes (scenarios) by the control means. The predetermined control process is, for example, a process for controlling a device or a function to be controlled based on information acquired from an utterance, a process for controlling a response by voice or screen display to a user, and the like. At this time, according to the present invention, for example, even if the user's natural utterance is input, since the keyword is appropriately grasped and the final speech candidate group is determined, the predetermined control processing is performed by the user. It can be determined and executed appropriately according to the intention.

  The control means considers the state of the system (for example, a vehicle) in which the speech recognition device is mounted, the state of the user, the state of the device or function to be controlled, and the like together with the final speech candidate group. It is also possible to determine and execute a predetermined control process. In addition, a storage unit is provided for storing a user's dialog history and device state changes, and the control unit performs predetermined control processing in consideration of the dialog history and state changes together with the final voice candidate group. It is also possible to decide.

Next, the speech recognition method of the present invention includes a text conversion step of converting the speech into text expressed by a word string by recognizing the input speech, and the text converted in the text conversion step. A process of analyzing the text using a word string score calculated based on the characteristics of the word and the word string is executed, and a result of the process is output as a first voice candidate group together with a score of each voice candidate. A process of analyzing the text using a word score calculated based on the feature of the word included in the text converted in the text analysis step and the text conversion step, and the result of the process is a second result As a speech candidate group, a second syntactic analysis step that is output together with a score of each speech candidate, and a first that takes the highest score among speech candidates included in the first speech candidate group The comparison judgment step for comparing the voice candidate with the second voice candidate having the highest score among the voice candidates included in the second voice candidate group and judging whether or not they match, and the comparison judgment step If it is determined that they do not match, the score determination step for increasing the highest score of the second speech candidate by a predetermined value, and the first speech candidate group and the second speech candidate group, based on the score of each speech candidate And a speech candidate group determination step for determining a final speech candidate group.

  According to the speech recognition method of the present invention, as described with respect to the speech recognition device of the present invention, the score of the second speech candidate that takes the highest score in the second speech candidate group is increased. There is a high possibility that the speech candidates to be included are higher in the final speech candidate group. Therefore, according to this speech recognition method, even if the user's natural utterance is input, the keyword is appropriately grasped, so that the recognition accuracy of the utterance can be improved.

  Next, the speech recognition program of the present invention is included in a text conversion process for recognizing input speech to convert the speech into text expressed by a word string, and text converted by the text conversion processing. First text analysis that analyzes the text using a word string score calculated based on the characteristics of the word and the word string, and outputs the result of the analysis as a first voice candidate group together with the score of each voice candidate Each of the speech candidates by analyzing the text using a word score calculated based on the processing and a feature of the word included in the text converted by the text conversion processing, and using the result of the analysis as a second speech candidate group A second syntactic analysis process that is output together with the scores of the first voice candidate group, the first voice candidate group, and the second voice candidate group to determine whether there is a matching voice candidate. Then, by the comparison determination process, the first speech candidate that has the highest score among the speech candidates included in the first speech candidate group and the highest score among the speech candidates that are included in the second speech candidate group. When it is determined that the second speech candidate does not match, the score determination process for increasing the highest score of the second speech candidate by a predetermined value, the first speech candidate group, and the second speech candidate group The computer has a function of causing a computer to execute a speech candidate group determination process for determining a final speech candidate group based on the score of each speech candidate.

  In this case, it is possible to cause the computer to execute processing that can achieve the effects described in regard to the speech recognition apparatus of the present invention.

[First Embodiment]
As shown in FIG. 1, the speech recognition apparatus according to the first embodiment of the present invention includes a speech dialogue unit 1 and is mounted on a vehicle 10. The voice interaction unit 1 is connected to a microphone 2 to which an utterance is input from a driver of the vehicle 10, and to a vehicle state detection unit 3 that detects the state of the vehicle 10. In addition, a speaker 4 that outputs a response to the driver and a display 5 that displays to the driver are connected to the voice interaction unit 1. Furthermore, a plurality of devices 6 a to 6 c that can be operated by the driver by voice or the like are connected to the voice interaction unit 1.

  The microphone 2 is for inputting the voice of the driver of the vehicle 10 and is installed at a predetermined position in the vehicle. For example, when a voice switch is instructed by a talk switch, the microphone 2 acquires the input voice as the driver's utterance. The talk switch is an ON / OFF switch that is operated by the driver of the vehicle 10, and is commanded to start voice input when pressed by being pressed.

  The vehicle state detection unit 3 is a sensor or the like that detects the state of the vehicle 10. The state of the vehicle 10 includes, for example, the traveling state of the vehicle 10 such as speed and acceleration / deceleration, traveling environment information such as the position of the vehicle 10 and the traveling road, and the equipment (wiper, winker, navigation system 6a, The operation state of the audio 6b or the like, or the vehicle interior state such as the vehicle interior temperature of the vehicle 10. Specifically, for example, as a sensor that detects the traveling state of the vehicle 10, a vehicle speed sensor that detects the traveling speed (vehicle speed) of the vehicle 10, a yaw rate sensor that detects the yaw rate of the vehicle 10, and a brake operation (brake pedal) of the vehicle 10 And a brake sensor for detecting whether or not the engine is operated. Further, as the state of the vehicle 10, the state of the driver of the vehicle 10 (perspiration of the palm of the driver, driving load, etc.) may be detected.

  The speaker 4 outputs a response (voice guide) to the driver of the vehicle 10. In addition, as this speaker 4, the speaker which the below-mentioned audio 6a has can be used.

  The display 5 is, for example, a HUD (head-up display) that displays information such as an image on the front window of the vehicle 10, a display that is provided integrally with a meter that displays a traveling state such as the vehicle speed of the vehicle 10, or It is the display etc. with which the navigation system 6b was equipped. The display of the navigation system 6b is a touch panel 24 in which a touch switch is incorporated.

  The devices 6a to 6c are specifically an audio 6a, a navigation system 6b, and an air conditioner 6c installed in the vehicle 10. In each of the devices 6a to 6c, controllable components (device, content, etc.), functions, operations, and the like are determined in advance.

  For example, the audio 6a includes “CD”, “MP3”, “radio”, “speaker”, and the like as devices. Further, there is a “volume” as a function of the audio 6a. The operation of the audio 6a includes “change”, “on”, “off”, and the like. Furthermore, “CD” and “MP3” operations include “play” and “stop”. “Radio” functions include “channel selection”. In addition, the “volume” operation includes “up”, “down”, and the like.

  Further, for example, the navigation system 6b includes “screen display”, “route guidance”, “POI search”, and the like as contents. Further, the “screen display” operation includes “change”, “enlarge”, “reduce”, and the like. “Route guidance” is a function of guiding to a destination by voice guidance or the like, and “POI search” is a function of searching for a destination such as a restaurant or a hotel.

  Further, for example, the air conditioner 6c has “air volume”, “set temperature”, and the like as its functions. The operation of the air conditioner 6c includes “on” and “off”. Further, “change”, “increase”, “decrease” and the like are included in the operations of “air volume” and “set temperature”.

  These devices 6a to 6c are controlled by designating information (device and function types, operation contents, etc.) for controlling the target. The devices, contents, and functions of the devices 6a to 6c to be controlled are classified into a plurality of domains. “Domain” means classification according to the category of the recognition target, and specifically represents the device or function that is the control target. The domain can be specified hierarchically, for example, such that the “audio” domain is classified into the “CD” and “radio” domains below it.

  Although not shown in detail, the voice interaction unit 1 is composed of an electronic circuit including an A / D conversion circuit, a microcomputer (CPU, RAM, ROM), etc., and the output (analog signal) of the microphone 2 is A / D. It is converted into a digital signal and input through a conversion circuit. Then, the voice interaction unit 1 recognizes the utterance input from the driver based on the input data, and performs dialogue with the driver via the speaker 4 or the display 5 based on the recognition result. A process for presenting information to the driver, a process for controlling the devices 6a to 6c, and the like are executed. These processes are realized by the voice interaction unit 1 executing a program pre-installed in the memory of the voice interaction unit 1. This program includes the speech recognition program of the present invention. The program may be stored in a memory via a recording medium such as a CD-ROM, distributed or broadcast from an external server via a network or an artificial satellite, and received by a communication device mounted on the vehicle 10. In addition, it may be stored in a memory.

  More specifically, the voice interaction unit 1 is recognized as a function realized by the above program, a text conversion unit 11 that recognizes an input voice using the acoustic model 15 and the language model 16 and outputs it as text. And a syntax analysis unit 12 for understanding the meaning of the utterance from the text using the syntax model 17. The voice interaction unit 1 also determines a scenario using the scenario database 18 based on the recognition result of the utterance, and responds to the driver, controls the device, and the like, and the voice to be output to the driver And a speech synthesizer 14 that synthesizes a response by using a phoneme model 21.

  The acoustic model 15, the language model 16, the syntax model 17, the scenario database 18, and the phoneme model 19 are recording media (databases) such as a CD-ROM, a DVD, and an HDD in which data is recorded.

  The text conversion unit 11 constitutes the text conversion means of the present invention. In addition, the syntax analysis unit 12 constitutes a first syntax analysis unit 31, a second syntax analysis unit 32, a speech candidate group determination unit 33, and a comparison determination unit 34 according to the present invention. Further, the scenario control unit 13 constitutes the control means of the present invention.

  The text conversion unit 11 extracts the feature vector by performing frequency analysis on the waveform data indicating the speech sound input to the microphone 2. Then, the text conversion unit 11 executes “text conversion processing” for recognizing the input voice based on the extracted feature vector and outputting it as text expressed in a word string. This text conversion process is executed by comprehensively determining the acoustic features and linguistic features of the input speech using a probability statistical method as described below.

  That is, the text conversion unit 11 first evaluates the likelihood of the pronunciation data according to the extracted feature vector using the acoustic model 15 (hereinafter, this likelihood is referred to as “acoustic score” as appropriate). Pronunciation data is determined based on the acoustic score. Further, the text conversion unit 11 uses the language model 16 to evaluate the likelihood of the text expressed by the word string corresponding to the determined pronunciation data (hereinafter, this likelihood is referred to as “language score” as appropriate). The text is determined based on the language score. Furthermore, the text conversion unit 11 determines the certainty of text conversion for all the determined texts based on the acoustic score and language score of the text (hereinafter, this certainty is referred to as “text conversion score” as appropriate). calculate. Then, the text conversion unit 11 outputs a text whose text conversion score satisfies a predetermined condition as a recognized text (Recognized Text).

  The syntax analysis unit 12 executes “syntax analysis processing” for understanding the meaning of the input utterance from the text recognized by the text conversion unit 11 using the syntax model 17. This parsing process is executed by analyzing a relationship (syntax) between words in the text recognized by the text conversion unit 11 using a probabilistic statistical method as described below.

  That is, the parsing unit 12 evaluates the likelihood of the recognized text (hereinafter, this likelihood is appropriately referred to as a “parsing score”), and the meaning of the recognized text is determined based on the parsing score. The text classified into the class corresponding to is determined. Then, the parsing unit 12 outputs a text classified by the classification (Categorized Text) such that the parsing score satisfies a predetermined condition as a recognition result of the input utterance together with the parsing score. A class corresponds to a classification according to a category representing a control target or control content, such as the domain described above. For example, when the recognized text is “change setting”, “change setting”, “change setting”, “change setting”, the classified texts are all {Setup}.

  The scenario control unit 13 uses the data recorded in the scenario database 18 on the basis of at least the recognition result output from the syntax analysis unit 12 and the state of the vehicle 10 acquired from the vehicle state detection unit 3. Determine the response output to the user and the device control scenario. In the scenario database 18, a plurality of scenarios for response output and device control are recorded in advance together with the utterance recognition result and vehicle condition. Then, the scenario control unit 13 executes a process for controlling a response by voice or image display or a process for controlling a device according to the determined scenario. Specifically, for example, in the case of a voice response, the scenario control unit 13 informs the user of the content of the response to be output (response sentence for prompting the driver's next utterance, completion of the operation, etc. Response sentence) and the speed and volume when outputting the response.

  The speech synthesizer 14 synthesizes speech using the phoneme model 19 in accordance with the response sentence determined by the scenario control unit 13 and outputs it as waveform data indicating the speech. The voice is synthesized using a process such as TTS (Text to Speech). Specifically, the speech synthesis unit 14 normalizes the text of the response sentence determined by the scenario control unit 13 into an expression suitable for speech output, and converts each word of the normalized text into pronunciation data. Then, the speech synthesizer 14 determines a feature vector from the pronunciation data using the phoneme model 19 and performs filtering on the feature vector to convert it into waveform data. This waveform data is output from the speaker 4 as sound.

  In the acoustic model 15, data indicating a probabilistic correspondence between the feature vector and the pronunciation data is recorded. Specifically, in the acoustic model 15, a plurality of HMMs (Hidden Markov Models) prepared for each recognition unit (phoneme, morpheme, word, etc.) are recorded as data. The HMM is a statistical signal source model that expresses speech as a connection of stationary signal sources (states) and expresses a time series as a transition probability from one state to the next state. With the HMM, it is possible to represent the acoustic features of speech that varies in time series with a simple probability model. Parameters such as transition probabilities of the HMM are determined in advance by giving corresponding learning speech data for learning. The phoneme model 19 also records the same HMM as the acoustic model 15 for determining the feature vector from the pronunciation data.

  In the language model 16, data indicating the appearance probability and connection probability of a word to be recognized is recorded together with pronunciation data and text of the word. The word that is the recognition target is predetermined as a word that may be used in the utterance for controlling the target. Data such as word appearance probabilities and connection probabilities are statistically created by analyzing a large amount of learning text corpus. Further, the appearance probability of a word is calculated based on, for example, the appearance frequency of the word in the learning text corpus.

  As the language model 16, for example, an N-gram language model expressed by the probability that specific N words appear successively is used. In the present embodiment, N-grams corresponding to the number of words included in the input utterance are used for the language model 16. Specifically, the language model 16 uses N-grams in which the value of N is equal to or less than the number of words included in the pronunciation data. For example, when the number of words included in the pronunciation data is 2, a unigram (Uni-gram, N = 1) represented by the appearance probability of one word, and the occurrence probability of a sequence of two words (for the preceding one word) A bigram (Bi-gram, N = 2) expressed by a conditional appearance probability) is used.

  Furthermore, in the language model 16, N gram can be used by limiting the value of N to a predetermined upper limit value. As the predetermined upper limit value, for example, a predetermined value (for example, N = 2), a value that is sequentially set so that the processing time of the text conversion processing for the input utterance is within a predetermined time, or the like is used. be able to. For example, when N-grams are used with N = 2 as the upper limit, only unigrams and bigrams are used even when the number of words included in the pronunciation data is greater than two. Thereby, it is possible to prevent the calculation cost of the text conversion processing from becoming excessive, and to output a response with an appropriate response time to the driver's utterance.

  In the syntax model (Parser Model) 17, data indicating the appearance probability and connection probability of a word to be recognized is recorded together with the text and class of the word. For example, an N-gram language model is used for the syntax model 17 in the same manner as the language model 16. In the present embodiment, specifically, the syntax model 17 uses N-grams equal to or less than the number of words included in the text in which the value of N is recognized with N = 3 as the upper limit. That is, in the syntax model 17, a trigram (Tri-gram, N = 3) represented by the occurrence probability (conditional appearance probability of the preceding two words) of a unigram, a bigram, and a sequence of three words is used. . The upper limit value may be other than 3, and can be arbitrarily set. Moreover, it is good also as what uses N gram below the number of words contained in the text by which the value of N was recognized, without restrict | limiting to an upper limit.

  As shown in FIG. 2, the language model 16 and the syntax model 17 are created by being classified for each type of domain. In the example of FIG. 2, there are eight types of domains: {Audio, Climate, Passenger Climate, POI, Ambiguous, Navigation, Clock, Help}. {Audio} indicates that the control target is the audio 6a. {Climate} indicates that the controlled object is the air conditioner 6c. {Passenger Climate} indicates that the control target is the air conditioner 6c of the passenger seat. {POI} indicates that the control target is the POI search function of the navigation system 6b. {Navigation} indicates that the control target is a function such as route guidance or map operation of the navigation system 6b. {Clock} indicates that the controlled object is a clock function. {Help} indicates that the control target is a help function for knowing how to operate the devices 6a to 6c and the speech recognition apparatus. {Ambiguous} indicates that the control target is unknown.

  Next, the operation (voice dialogue processing) of the voice recognition device of this embodiment will be described. As shown in FIG. 3, first, in STEP 1, an utterance for controlling an object is input to the microphone 2 from the driver of the vehicle 10. Specifically, the driver turns on the talk switch to instruct the start of utterance input, and inputs sound into the microphone 2.

  Next, in STEP 2, the voice interaction unit 1 executes text conversion processing for recognizing the input voice and outputting it as text.

  First, the voice interaction unit 1 performs A / D conversion on the voice input to the microphone 2 to obtain waveform data indicating the voice. Next, the voice interaction unit 1 performs frequency analysis on the waveform data indicating the voice and extracts a feature vector. Thereby, the waveform data indicating the voice is subjected to filter processing by, for example, a technique of short-time spectrum analysis, and converted into a time series of feature vectors. This feature vector is obtained by extracting a feature amount of a speech spectrum at each time, and generally has 10 to 100 dimensions (for example, 39 dimensions), such as LPC mel cepstrum (Linear Predictive Coding) coefficients, etc. Is used.

  Next, the voice interaction unit 1 evaluates the likelihood (acoustic score) of the feature vector for each of the plurality of HMMs recorded in the acoustic model 15 with respect to the extracted feature vector. Then, the voice interaction unit 1 determines pronunciation data corresponding to the HMM having a high acoustic score among the plurality of HMMs. Thus, for example, when an utterance “Chitose” is input, pronunciation data “ti-to-se” is obtained together with the acoustic score from the waveform data of the speech. At this time, for example, when the utterance "mark set" is input, the pronunciation data "ma-a-ku-se-t-to" and "ma-a-ku-ri-su-to" The pronunciation data having a high degree of acoustic similarity are obtained together with the acoustic score.

  Next, the voice interaction unit 1 determines the text expressed by the word string from the determined pronunciation data based on the language score of the text. At this time, when a plurality of pronunciation data are determined, text is determined for each pronunciation data. For example, when the domain type is determined based on, for example, an input from the driver to the touch panel 24, the voice interaction unit 1 is classified into the determined type domain in the language model 16. The process of determining the following text may be performed using only the data of the part that has been set.

  First, the voice interaction unit 1 uses the language model 16 to determine text from pronunciation data. Specifically, first, the voice interaction unit 1 compares the determined pronunciation data with the pronunciation data recorded in the language model 16 and extracts words having a high degree of similarity.

  Next, the voice interaction unit 1 calculates the language score of the extracted word using an N-gram according to the number of words included in the pronunciation data. Then, the voice interaction unit 1 determines the text for which the calculated language score satisfies a predetermined condition (for example, a predetermined value or more) for each word in the pronunciation data. For example, as shown in FIG. 4, when the input utterance is “Set the station ninety nine point three FM.”, The text corresponding to the pronunciation data determined from this utterance is “set the station ninety nine”. “point three FM” is determined.

  At this time, in the unigram, the appearance probabilities a1 to a8 of “set”, “the”... “FM” are given. In the bigram, occurrence probabilities b1 to b7 of two words “set the”, “the station”,..., “Three FM” are given. Similarly, for N = 3 to 8, N word occurrence probabilities c1 to c6, d1 to d5, e1 to e4, f1 to f3, g1 to g2, and h1 are given. For example, the language score of the text “ninety” is obtained from N grams of N = 1 to 4 according to the number of words 4 including the word “ninety” included in the pronunciation data and the word preceding the word. It is calculated based on a4, b3, c2, and d1.

  In this way, by using a technique (dictation) that transcribes input utterances as text using a probabilistic language model for each word, the driver's natural utterances are not limited to utterances of predetermined phrases. Utterances can be recognized.

  Next, the speech interaction unit 1 calculates the certainty of text conversion (text conversion score) that is a weighted sum of the acoustic score and the language score for all texts determined using the language model 16. As the weighting factor, for example, a value predetermined experimentally is used.

  Next, the voice interaction unit 1 determines and outputs a text represented by a word string whose calculated text conversion score satisfies a predetermined condition as a recognized text. The predetermined condition is determined in advance as, for example, text having the highest text conversion score, text having a text conversion score from the top to a predetermined rank, or text having a text conversion score of a predetermined value or more.

  Next, in STEP 3, the voice interaction unit 1 executes a parsing process for understanding the meaning of the utterance from the recognized text.

  First, the spoken dialogue unit 1 uses the syntax model 17 to determine the classified text from the recognized text. Specifically, first, the voice interaction unit 1 calculates the likelihood for each domain for each word using the entire syntax model 17. Next, the voice interaction unit 1 determines a domain based on the likelihood calculated for each word. Next, the voice interaction unit 1 calculates the likelihood for each class set for each word using the data of the portion of the syntax model 17 classified into the determined domain type. Then, a class set (classified text) is determined based on the likelihood (word score) calculated for each word. Note that the process of determining a class set for each word corresponds to the process of the second syntax analysis unit 32, and the determined set of one or more classes corresponds to a second speech candidate group. Here, the “speech candidate” indicates a command candidate indicating a control target or control content obtained by analyzing text recognized from the input speech.

  Similarly, the voice interaction unit 1 calculates the likelihood of each domain in two words for each of the two word strings included in the recognized text, and determines the domain in the two words based on the likelihood. Furthermore, the voice interaction unit 1 calculates the likelihood (two-word score) of each class set in two words, and determines a class set (class-categorized text) in two words based on the two-word score. . Similarly, the voice interaction unit 1 calculates the likelihood of each domain in the three words for each of the three word strings included in the recognized text, and determines the domain in the three words based on the likelihood. . Furthermore, the voice interaction unit 1 calculates the likelihood (three-word score) of each class set in three words, and determines the class set (class-categorized text) in three words based on the three-word score. .

  Next, the voice interaction unit 1 is based on each class set determined by 1 word, 2 words, and 3 words and the score of the class set (1 word score, 2 word score, 3 word score), The likelihood (parse score) of each class set in the entire recognized text is calculated. Then, the voice interaction unit 1 determines a class set (classified text) in the entire recognized text based on the parsing score. Note that the process of determining a class set for the entire text corresponds to the process of the first syntax analysis unit 31, and the determined set of one or more classes corresponds to a first speech candidate group.

  Here, using the example shown in FIG. 5, processing for determining text classified by using the syntax model 17 will be described. In the example of FIG. 5, the recognized text is “AC on floor to defrost”.

  At this time, the likelihood of each domain in one word is calculated for each of “AC”, “on”... “Defrost” as a unigram using the entire syntax model 17. Then, a domain in one word is determined based on the likelihood. For example, the first (highest likelihood) domain is determined as {Climate} for “AC”, {Ambiguous} for “on”, and {Climate} for “defrost”.

  Further, using the data of the portion of the syntax model 17 classified into the determined domain type, the likelihood for each class set in one word for “AC” “on”. Are calculated respectively. Then, a class set in one word is determined based on the likelihood. For example, with respect to “AC”, the class set having the highest rank (highest likelihood) is determined as {Climate_ACOnOff_On}, and the likelihood (word score) i1 for this class set is obtained. Similarly, for “on”... “Defrost”, a class set is determined, and likelihoods (word scores) i2 to i5 for the class set are obtained. Thereby, a speech candidate list for each word as shown in FIG. 6A is obtained as the second speech candidate group.

  Similarly, for each of “AC on”, “on floor”... “To defrost” in the bigram, the likelihood of each domain in two words is calculated, and the domain in two words is determined based on the likelihood. Then, a class set in two words and its likelihood (two-word score) j1 to j4 are determined. Similarly, in the trigram, for each of “AC on floor”, “on floor to”, and “floor to defrost”, the likelihood of each domain in three words is calculated, and the domain in three words is calculated based on the likelihood. Is determined. Then, a class set in three words and its likelihood (three word score) k1 to k3 are determined.

  Next, for each class set determined by one word, two words, and three words, for example, the sum of the word scores i1 to i5, the two word scores j1 to j4, and the three word scores k1 to k3 of each class set is The likelihood (syntactic analysis score) for each set of classes in the entire text is calculated. For example, the parsing score for {Climate_Fan-Vent_Floor} is i3 + j2 + j3 + k1 + k2 = 0.8. For example, the parsing score for {Climate_ACOnOff_On} is i1 + j1 = 0.6. For example, the parsing score for {Climate_Defrost_Front} is i5 + j4 = 0.5. Then, based on the calculated parsing score, a class set (classified text) for the entire text is determined. Thereby, a speech candidate list of word strings as shown in FIG. 6A is obtained as the first speech candidate group.

  Next, in STEP 4, the voice interaction unit 1 takes the first voice candidate that has the highest score among the voice candidates included in the first voice candidate group and the highest score among the second voice candidates. The two voice candidates are compared to determine whether or not they match. In the example shown in FIG. 6, the first speech candidate {Climate_Fan-Vent_Floor} in the speech candidate list of the word string is compared with the first speech candidate {Climate_Defrost_Front} in the speech candidate list for each word. It is determined whether or not to do so.

  If the determination result in STEP 4 is YES (matches), the process proceeds to STEP 6 as it is. If the determination result in STEP 4 is NO (does not match), the process proceeds to STEP 5 and the voice interaction unit 1 increases the highest score of the second voice candidate by a predetermined value.

  In the example shown in FIG. 6, the first speech candidate {Climate_Fan-Vent_Floor} in the word sequence speech candidate list does not match the first speech candidate {Climate_Defrost_Front} in the speech candidate list for each word. The determination result is NO. In this case, the first speech candidate {Climate_Defrost_Front} in the speech candidate list for each word is a speech candidate including the meaning of the keyword “defrost” included in the input speech. At this time, the process proceeds to STEP 5 and, as shown in FIG. 6B, the score (word score) of the first speech candidate {Climate_Defrost_Front} in the speech candidate list for each word is increased by 0.2. Reflecting this score, the likelihood (syntactic analysis score) for each class set is calculated for the entire text. For example, the parsing score is calculated as the sum of the word score determined in STEP 5, the two-word score, and the three-word score, similarly to the word string score. Then, based on the calculated parsing score, an all speech candidate list as shown in FIG. 6B is obtained. As a result, the third speech candidate {Climate_Defrost_Front} in the word candidate speech list of the word string in FIG. 6A is increased by 0.2 in the all speech candidate list in FIG. It becomes the second speech candidate. That is, the score of the speech candidate {Climate_Defrost_Front} including the meaning of the keyword is increased and the ranking is increased. Then, the process proceeds to STEP6.

  Next, in STEP 6, the voice interaction unit 1 selects one or a plurality of voice candidates whose parsed scores satisfy a predetermined condition from the voice candidates included in the all voice candidate list. It is determined as a recognition result (final speech candidate group) and is output together with the certainty of the recognition result (syntactic analysis score of each speech candidate). The predetermined condition is determined in advance, for example, as a speech candidate having the highest syntax analysis score, a speech candidate having a syntax analysis score from a higher rank to a predetermined rank, or a speech candidate having a syntax analysis score of a predetermined value or more.

  At this time, since the score of the speech candidate including the meaning of the keyword is increased in STEP 5 and the rank is higher, the possibility that the speech candidate is included in the final speech candidate group is high. For example, when the utterance “AC on floor to defrost” is input as described above, if the predetermined condition is a speech candidate with a parsing score up to the second place, the final speech in FIG. As shown in the candidate list, {Climate_Fan-Vent_Floor} {Climate_Defrost_Front} is output together with the score as a recognition result. That is, the speech candidate {Climate_Defrost_Front} including the keyword meaning is included in the final speech candidate group. This process corresponds to the process of the speech candidate group determination unit 33.

  Next, in STEP 7, the voice interaction unit 1 detects the state of the vehicle 10 (the traveling state of the vehicle 10, the state of the device mounted on the vehicle 10, the state of the driver of the vehicle 10) detected by the vehicle state detection unit 3. Etc.) is obtained.

  Next, in STEP 8, the voice interaction unit 1 uses the scenario database 18 based on the final voice candidate group (utterance recognition result) determined in STEP 6 and the state of the vehicle 10 detected in STEP 5. The scenario for responding to the driver and controlling the device is determined. At this time, since the keyword is appropriately grasped and the final speech candidate group is determined in STEP 6 and the recognition accuracy of the recognition result of the driver's utterance is improved, the scenario is appropriate according to the driver's intention. To be determined.

  First, the voice interaction unit 1 acquires information for controlling a target from the utterance recognition result and the state of the vehicle 10. As shown in FIG. 7, the voice interaction unit 1 is provided with a plurality of forms for storing information for controlling an object. Each form has a predetermined number of slots corresponding to the class of information required. For example, “Plot a route” and “Traffic info.” Are provided as forms for storing information for controlling the navigation system 6b, and “Climate control” is provided as a form for storing information for controlling the air conditioner 6c. And the like. The form “Plot a route” is provided with four slots “From”, “To”, “Request”, and “via”.

  The voice dialogue unit 1 inputs a value into a slot of the corresponding form from the recognition result of each utterance in the dialogue with the driver and the state of the vehicle 10. Specifically, the voice interaction unit 1 determines a voice candidate from the final voice candidate group in consideration of the state of the vehicle 10, and inputs a value into a slot of a form corresponding to this voice candidate. At the same time, the certainty factor (degree of confidence of the value input to the form) for each form is calculated and recorded on the form. For example, the certainty factor of the form is calculated based on the certainty factor of the recognition result of each utterance and the filling degree of the slot of each form. For example, as shown in FIG. 8, when an utterance “Guide to the shortest route to Chitose Airport” is input from the driver, three slots “From” “To” “ The values “here”, “Chitose Airport” and “shortest” are entered in “Request”. Further, the calculated confidence factor 80 of the form is recorded in “Score” of the form “Plot a route”.

  Next, the voice interaction unit 1 selects a form to be used for actual control processing based on the certainty of the form and the state of the vehicle 10 detected in STEP 7. Then, based on the selected form, the scenario is determined using the data stored in the scenario database 18. As shown in FIG. 8, in the scenario database 18, for example, response sentences (prompts) to be output to the driver are classified and stored for each slot filling level and level. Note that the level is a value set based on, for example, the certainty of the form, the state of the vehicle 10 (the traveling state of the vehicle 10, the state of the driver, and the like).

  For example, when there is an empty slot (a slot in which no value is input) in the selected form, a scenario is determined for outputting a response sentence that prompts the driver to input an empty slot in the form. At this time, an appropriate response sentence that prompts the driver to speak next time is determined according to the level, that is, taking into account the certainty of the form and the state of the vehicle 10. For example, in a state where the driving load is considered to be high according to the driving load of the driver, a response sentence in which the number of slots for prompting input is set to be small is determined. Then, by prompting the user's next utterance by outputting the response sentence determined in this way, an efficient dialogue is performed.

  In the example shown in FIG. 8, values are input to the first to third slots “From”, “To”, and “Request” of the form “Plot a route”, and values are input to the fourth slot “via”. Not. Further, level = 2 is set. At this time, the response sentence “<To> <Request> is set” is selected from the scenario database 18, and the content of the response sentence “High-speed priority setting is set for Chitose Airport” is determined.

  In addition, for example, when all slots in the selected form are all filled (values are input), a response sentence that confirms the contents (for example, the input value of each slot is notified to the driver) Response scenario) is determined.

  Next, in STEP 9, the voice interaction unit 1 determines whether or not the dialogue with the driver has ended based on the determined scenario. If the determination result in STEP 9 is NO, the process proceeds to STEP 10, where the voice interaction unit 1 synthesizes speech according to the contents of the determined response sentence and the conditions for outputting the response sentence. Then, in STEP 10, the generated response text (such as a response text prompting the driver to speak next time) is output from the speaker 4.

  Next, returning to STEP 1, the second utterance is input from the driver. Next, in STEPs 2 to 8, the voice interaction unit 1 executes processing for the second utterance as in the first utterance. At this time, in STEP 2, for example, when the type of the domain is determined based on the recognition result of the previous utterance from the driver, the voice interaction unit 1 determines the determined type of the language model 16. The text may be determined using only the data of the portion classified into the domain. Next, in STEP 9, the voice interaction unit 1 determines whether or not the dialogue with the driver has ended. When the determination result in STEP 9 is NO, the voice interaction unit 1 executes the processing in STEPs 10 to 11 as in the first utterance.

  Thereafter, the processing similar to STEP 1 to STEP 11 for the second utterance described above is repeated until the determination result of STEP 9 is YES.

  If the determination result in STEP 9 is YES, the process proceeds to STEP 12 and the voice interaction unit 1 synthesizes the voice of the determined response sentence (such as a response sentence that informs the contents of device control). Next, a response sentence is output from the speaker 4 in STEP13. Next, in STEP 14, the voice interaction unit 1 controls the device based on the determined scenario, and ends the voice interaction process.

Through the above processing, keywords are properly grasped and the recognition accuracy of the utterance is improved, so that the device is controlled through an efficient dialogue.
[Second Embodiment]
Next, the speech recognition apparatus according to the second embodiment of the present invention will be described. The present embodiment is different from the first embodiment only in the process of determining the final speech candidate group (utterance recognition result) of STEP 6 in the speech dialogue processing. Since the configuration of this embodiment is the same as that of the first embodiment, the same reference numerals are given to the same configuration, and the description thereof is omitted below.

  The operation (voice dialogue processing) of the voice recognition device of the present embodiment will be described using the examples shown in FIGS. In the example of FIGS. 9 to 11, the recognized text is “Tune the next disc”. At this time, in STEP 3, as in the first embodiment, a class set of one word, two words, and three words and a score of the class set are calculated. FIG. 9 shows a set of the first class (highest likelihood) class. FIG. 10 also shows the second class set.

  At this time, for the first class set shown in FIG. 9, as in STEP 4 of the first embodiment, the first speech candidate {Audio_CD_Nextpre_Next} in the speech candidate list of the word string and the speech for each word The first candidate audio candidate {Audio_CD_DiscNo} in the candidate list is compared to determine whether or not they match. Since {Audio_CD_Nextpre_Next} and {Audio_CD_DiscNo} do not match, the determination result in STEP 4 is NO. In this case, the first speech candidate {Audio_CD_DiscNo} in the speech candidate list for each word is a speech candidate including the meaning of the keyword “disc” included in the input speech. At this time, the process proceeds to STEP 5 and, as shown in FIG. 9, the score of {Audio_CD_DiscNo} in the speech candidate list for each word is increased, and the parsing score is calculated to reflect this score. A candidate list is obtained. Accordingly, the third speech candidate {Audio_CD_DiscNo} in the speech candidate list of the word string is increased in score in the first speech candidate list and becomes the second speech candidate in the speech candidate list. That is, the score of the audio candidate {Audio_CD_DiscNo} including the meaning of the keyword is increased and the ranking is increased.

  For the second class set shown in FIG. 10, the word string score is directly used as the syntax analysis score, and the speech candidate list of the word string is directly used as the second highest speech candidate list. Then, the process proceeds to STEP6.

  Next, in STEP 6, as shown in FIG. 11, the voice interaction unit 1 multiplies the score of each voice candidate by a predetermined value (for example, 0.9) less than 1 for the second highest voice candidate list. Then, a final speech candidate list is determined based on the score of each speech candidate from the speech candidate group in which the first speech candidate list and the second speech candidate list are combined. Other operations are the same as those in the first embodiment.

According to the speech recognition apparatus of the present embodiment, as in the first embodiment, keywords are appropriately grasped and speech recognition accuracy is improved, so that the device is controlled through efficient dialogue.
[Third Embodiment]
Next, a speech recognition apparatus according to a third embodiment of the present invention will be described. Note that the present embodiment is different from the first embodiment only in the process of determining the final speech candidate group (speech recognition result) of STEP 6 in the speech interaction process (the process of the speech candidate group determination unit 33). Since the configuration of this embodiment is the same as that of the first embodiment, the same reference numerals are given to the same configuration, and the description thereof is omitted below.

  In the speech recognition apparatus according to the present embodiment, the speech candidate group determination unit 33 drives a plurality of speech candidates included in the first and second speech candidate groups based on the scores of the speech candidates via the touch panel 24. And a voice guide is output through the speaker 4. Then, the voice candidate group determination unit 33 determines a final voice candidate based on a touch input for selecting at least one of voice candidates displayed on the touch panel 24 from the driver. The scenario control unit 13 and the speech synthesis unit 14 constitute notification means and selection means of the present invention.

  In the operation of the speech recognition apparatus of the present embodiment (speech dialogue processing), in STEP 6, first, the speech dialogue unit 1 starts from the speech candidates included in the first and second speech candidate groups based on the scores of the speech candidates. Then, a plurality of voice candidates to be displayed on the touch panel 24 are determined. Specifically, as shown in FIG. 6B, the speech analysis score of each speech candidate is set to a predetermined score from the entire speech candidate list in which the speech candidate score including the meaning of the keyword is increased and the ranking is high. A plurality of speech candidates that satisfy the condition are determined. The predetermined condition is determined in advance, for example, as a speech candidate having the highest syntax analysis score, a speech candidate having a syntax analysis score from a higher rank to a predetermined rank, or a speech candidate having a syntax analysis score of a predetermined value or more. Then, the voice interaction unit 1 presents information by displaying the plurality of voice candidates on the touch panel 24 so as to prompt the driver to select them. At this time, the voice interaction unit 1 synthesizes a response sentence (voice guide) that prompts the driver to make a selection by the voice synthesizer 14 and outputs it from the speaker 4.

  Thereby, the driver performs touch input for selecting at least one of the plurality of voice candidates displayed on the touch panel 24. At this time, since the score of the voice candidate including the meaning of the keyword is increased and the rank is higher in STEP 5, the voice candidate corresponding to the driver's utterance is included in the voice candidate displayed on the touch panel 24 with a high probability. Thus, the driver can select a voice candidate according to the intention of the utterance. Then, the voice interaction unit 1 determines the voice candidate group selected by the touch input as the final voice candidate group. Thereby, the speech candidate group that matches the utterance of the driver is determined as the final speech candidate group. Other operations are the same as those in the first embodiment.

  According to the speech recognition apparatus of the present embodiment, as in the first embodiment, keywords are appropriately grasped and speech recognition accuracy is improved, so that the device is controlled through efficient dialogue.

  In the first to third embodiments, the vehicle state detection unit 3 is provided, and the scenario control unit 13 determines the scenario according to the recognition result and the detected vehicle state. 3, the scenario control unit 13 may determine the control process only from the recognition result.

  In the first to third embodiments, the user who inputs the voice is the driver of the vehicle 10, but may be an occupant other than the driver.

  In the first to third embodiments, the voice recognition device is mounted on the vehicle 10, but may be mounted on a moving body other than the vehicle. Furthermore, the present invention is not limited to a mobile object, and can be applied to a system in which a user controls an object by speaking.

The functional block diagram of the speech recognition apparatus which is 1st Embodiment of this invention. FIG. 2 is an explanatory diagram illustrating a configuration of a language model and a syntax model of the speech recognition apparatus in FIG. 1. The flowchart which shows the whole operation | movement (voice dialogue process) of the speech recognition apparatus of FIG. Explanatory drawing which shows the text conversion process using the language model in the speech dialogue process of FIG. FIG. 4 is an explanatory diagram illustrating a syntax analysis process using a syntax model in the voice interaction process of FIG. 3. Explanatory drawing which shows the process which determines the final audio | voice candidate group in the audio | voice conversation process of FIG. Explanatory drawing which shows the form used for the process which determines the scenario in the voice dialog process of FIG. Explanatory drawing which shows the process which determines the scenario in the voice dialogue process of FIG. Explanatory drawing which shows the syntax analysis process in the speech dialogue process of the speech recognition apparatus of 2nd Embodiment of this invention. Explanatory drawing which shows the syntax analysis process in the voice dialogue process of FIG. Explanatory drawing which shows the process which determines the final audio | voice candidate group in the audio | voice dialog process of FIG.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Voice interaction unit, 2 ... Microphone, 3 ... Vehicle state detection part, 4 ... Speaker, 5 ... Display, 6a-6c ... Apparatus, 10 ... Vehicle, 11 ... Text conversion part, 12 ... Syntax analysis part, 13 ... Scenario Control unit, 14 ... synthesizer, 15 ... acoustic model, 16 ... language model, 17 ... syntax model, 18 ... scenario database, 19 ... phoneme model, 24 ... touch panel, 31 ... first syntax analysis means, 32 ... first 2 syntax analysis means, 33... Speech candidate group determination means, 34... Comparison judgment means.

Claims (6)

A text conversion means for recognizing the input speech to convert the speech into text represented by a word string;
Executing a process of analyzing the text using a word string score calculated based on a word and a characteristic of the word string included in the text converted by the text conversion unit, and the result of the process is a first speech candidate First parsing means for outputting together with the score of each speech candidate as a group;
Performing a process of analyzing the text using a word score calculated based on the characteristics of the words included in the text converted by the text conversion means, and using the result of the process as a second speech candidate group, A second parsing means for outputting together with the speech candidate score;
A speech candidate group determining means for determining a final speech candidate group from the first speech candidate group and the second speech candidate group based on the score of each speech candidate;
Comparing and determining means for comparing the first speech candidate group and the second speech candidate group to determine whether there is a matching speech candidate;
The comparison / judgment means obtains the highest score among the first speech candidates having the highest score among the speech candidates included in the first speech candidate group and the speech candidate included in the second speech candidate group. A speech recognition apparatus characterized in that, when it is determined that the second speech candidate to be taken does not match, the highest score of the second speech candidate is increased by a predetermined value. The speech recognition apparatus according to claim 1,
The speech candidate group determination means determines a speech candidate group whose score is within a predetermined rank among speech candidates included in the first and second speech candidate groups as the final speech candidate group. A voice recognition device characterized by the above. The speech recognition apparatus according to claim 1,
Informing means for informing a plurality of speech candidates included in the first and second speech candidate groups based on a score of each speech candidate, and selecting at least one of the plurality of speech candidates informed by the informing means A speech recognition apparatus, wherein the speech candidate group determination unit determines the speech candidate group selected by the selection unit as the final speech candidate group.   4. The speech recognition apparatus according to claim 1, further comprising a control unit that executes a predetermined control process based on at least a final speech candidate group determined by the speech candidate group determination unit. Voice recognition device. A text conversion step of converting the voice into a text represented by a word string by recognizing the input voice;
Executing a process of analyzing the text using a word string score calculated based on a word and a characteristic of the word string included in the text converted in the text conversion step, and using the result of the process as a first speech candidate A first parsing step for outputting together with a score for each speech candidate as a group;
Performing a process of analyzing the text using a word score calculated based on the characteristics of the words included in the text converted in the text conversion step, and using the result of the process as a second speech candidate group, A second parsing step for outputting together with the speech candidate score;
A first speech candidate that has the highest score among speech candidates included in the first speech candidate group, and a second speech candidate that has the highest score among speech candidates included in the second speech candidate group A comparison and determination step for determining whether or not they match,
A score determination step of increasing the highest score of the second speech candidate by a predetermined value when it is determined that they do not match in the comparison determination step;
A speech candidate group determining step of determining a final speech candidate group from the first speech candidate group and the second speech candidate group based on a score of each speech candidate; Recognition method. A text conversion process for recognizing the input speech to convert the speech into text represented by a word string;
Analyzing the text using a word string score calculated based on the characteristics of the word and the word string included in the text converted by the text conversion process, the result of the analysis as a first speech candidate group, A first parsing process to be output together with the speech candidate score;
Analyzing the text using a word score calculated based on the characteristics of the words included in the text converted by the text conversion process, and using the result of the analysis as a second speech candidate group, the score of each speech candidate A second parsing process to be output together with
A comparison determination process for comparing the first speech candidate group and the second speech candidate group to determine whether there is a matching speech candidate;
As a result of the comparison and determination process, the first speech candidate having the highest score among speech candidates included in the first speech candidate group and the highest score among speech candidates included in the second speech candidate group. A score determination process for increasing the highest score of the second speech candidate by a predetermined value when it is determined that the second speech candidate to be taken does not match,
A function of causing the computer to execute a speech candidate group determination process for determining a final speech candidate group based on a score of each speech candidate from the first speech candidate group and the second speech candidate group; A speech recognition program characterized by that.

Images