JP2008076811A - 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 accurately recognizing user's utterance, even when the user's utterance is ambiguous. <P>SOLUTION: The voice recognition device 1 comprises: an operation specifying means 31 which performs specifying processing for specifying an operation object and an operation content by classifying a recognition result regarding voice into a category of the operation object and the operation content which are set beforehand, in order to determine the operation content of the operation object, based on the recognition result regarding input voice; an ambiguous word detecting means 32 for detecting an ambiguous word which is not classified by the operation specifying means 31, from the recognition result regarding voice; a candidate extracting means 33 for extracting the operation object which may be able to be specified by the detected ambiguous word, as an operation object candidate. When the operation specifying means 31 can specify the operation content and cannot specify the operation object, the specifying processing is performed by replacing the ambiguous word detected by the ambiguous word detecting means 32, with the candidate extracted by the candidate extracting means 33 in the operation specifying means 31. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a speech recognition apparatus, a speech recognition method, and a speech recognition program that recognizes speech input by a user and acquires information for controlling an object based on the recognition result.

  In recent years, in a system that performs operations such as device control, a speech recognition device that recognizes a voice input by a user and acquires information necessary for device operation or the like has been used. In such a voice recognition device, a voice (speech) input by the user is recognized, and the user is urged to respond to the next utterance based on the recognized result, whereby a dialogue with the user is performed. . Information necessary for operating the device is acquired from the result of recognizing the dialogue with the user. At this time, for example, using a speech recognition dictionary in which a command to be recognized is registered in advance, the utterance is recognized by comparing the feature amount of the input utterance with the feature amount of the command registered in the speech recognition dictionary. Is done.

  Such a voice recognition device is used for a user to operate a plurality of devices such as an audio device, a navigation device, and an air conditioner mounted on the vehicle, for example. For example, when a function for increasing / decreasing the volume of an audio device is operated as an operation target, the user instructs by voice how to change the function (operation content), and is referred to as “up” or “down”. The operation content is instructed with words (indicators). The same applies to the case of operating the function of changing the radio reception station as the operation target, and the user must utter the instruction word “up” or “down” in order to increase or decrease the “reception frequency”. The operation content is indicated by.

  As described above, when there are a plurality of functions that operate with the same instruction word, it is not possible to identify which operation target function by uttering only such an instruction word. ). In the above example, there are operations to increase / decrease the volume of the audio device to be operated and operations to increase / decrease the reception frequency of the radio even if the word (indicator) indicates the function of increase / decrease. It is necessary to clearly indicate which frequency is to be increased or decreased. That is, it is necessary to say “volume up” to increase “volume” and “frequency up” to increase frequency. Also, when adjusting the temperature of an air conditioner, it is necessary to say the words “temperature up” or “temperature down” in order to raise or lower the set temperature.

  However, for example, when the word “up” (indicator) is used in a vehicle-mounted device, the “volume” increases if the audio device is activated, and the “temperature” increases if the air conditioner is activated. It is common to indicate the function of the inside device, and there can only be a volume, especially when no other device is operating. Also, if the word “up” is issued continuously when both the audio device and the air conditioner are operating, there is a high probability that it is an instruction for the volume of the audio device. In spite of such circumstances, it is bothersome and inconvenient for the user to always utter the function word and the instruction word together.

  As countermeasures, for example, use the "up" instruction word to increase the volume, use the "up" instruction to increase the frequency, and increase the air conditioner set temperature. It is also conceivable to set in advance to use different instruction words for each function, such as instructing with the word “cold”. However, even when the function to increase or decrease the set amount for each function is performed with almost the same operation instructions, it is difficult for the user to understand that it is determined to use different words (indicators) for each function. , Device operation becomes troublesome.

  Therefore, in Patent Document 1 below, a specific instruction word indicating the operation content of the function of the device is used in common for a plurality of functions, and these devices are all in operation. In addition, there has been proposed a voice recognition device operating device that can perform a desired operation by estimating a function desired by a user only by uttering an instruction word.

  This voice recognition device operating device identifies a function word for selecting a function of the device and a command word for instructing operation of the function from the voice recognized by the voice recognition unit, and voice input of the command word alone is performed. An identification unit that detects the fact, an operation instruction database that records the function of the device and an instruction word for operating the device, and whether or not the instruction word identified by the identification unit is a common instruction word for a plurality of functions The operable function detection unit that is detected from the operation instruction database, the usage frequency database that records the usage frequency corresponding to the instruction word, and the instruction word that is common to a plurality of functions in the operable function detection unit When detected, the usage frequency search unit that searches the usage frequency database for the most frequently used function among the plurality of functions, and the function of the most frequently used function searched by the usage frequency search unit. An operation word forming unit that adds words to input instruction words and outputs them as operation words, and a use frequency correction unit that corrects data in the use frequency database when an operation instruction signal for the device is output It is.

According to this apparatus, the function desired by the user is estimated based on the use frequency data, so that the user operates the combination of the function word indicating the function and the instruction word only by speaking the specific instruction word. As a word, the operation of the device can be performed.
JP 2001-34292 A

  However, the above-described voice recognition device operating device basically corresponds to a function that is used most frequently to a specific word that indicates operation contents that are commonly used for a plurality of functions. For this reason, it is not possible to cope with the case where the user instructs another operation due to the difference in the phrase using a specific word. Also, for example, when the word indicating the operation target in the input voice uttered by the user is an ambiguous word (ambiguous word) such as “this”, “it”, “that”, the ambiguous word is considered in consideration of the difference in the wording. It does not estimate what the word shows. Therefore, there is a problem that the function is not properly estimated and the usability of the device is not good.

  In view of the above circumstances, the present invention is applicable to a case where an ambiguous word whose operation target cannot be specified is included in the user's utterance, and voice recognition capable of estimating an appropriate operation for the user's utterance. An object is to provide an apparatus.

  The present invention relates to a speech recognition apparatus that determines an operation content of an operation target based on a recognition result of an input voice, and classifies the recognition result of the voice into a predetermined operation target and operation content type. The operation specifying means for performing the specifying process for specifying the operation target and the operation content, the ambiguous word detecting means for detecting the ambiguous word that cannot be classified by the operation specifying means from the recognition result of the voice, and the ambiguous word A candidate extraction unit that extracts an operation target that may be specified by an ambiguous word detected by the detection unit as an operation target candidate. The operation specifying unit is capable of specifying an operation content and specifying an operation target. If it is not possible to replace the ambiguous word detected by the ambiguous word detecting means with the candidate extracted by the candidate extracting means, the operation specifying means performs the specifying process. To.

  According to the present invention, for example, an utterance for manipulating an object is inputted by voice, and the voice is recognized as text expressed by a word string. Here, “text” is a significant syntax expressed in a string of words and having a predetermined meaning. Then, the operation target specifying means classifies the recognition result into the types of the operation target and the operation content by analyzing the relationship between words included in the recognition result for the voice, for example, and thereby the operation target and the operation content. Is identified. At this time, if the input voice is ambiguous, the operation content is specified, but the operation target may not be specified. For example, the operation specifying means is such that a word or a word string indicating an operation target in the input voice is a general instruction word such as “this”, “it”, “that”, or is not recognized correctly because the voice is unclear. This is a case of an ambiguous word that cannot be classified by. At this time, in the present invention, the ambiguous word is detected from the recognition result of the speech, the ambiguous word is replaced with the operation target candidate, and the identification process is performed by the operation specifying unit, whereby the content indicated by the ambiguous word included in the utterance is displayed. It is possible to appropriately estimate the operation target and the operation content.

  Therefore, according to the present invention, it is possible to cope with a case where an ambiguous word whose operation target cannot be specified is included in the user's utterance, and an appropriate operation can be performed by extracting candidates for specifying the operation target. Can be estimated.

  In a preferred embodiment of the present invention, the candidate extracting unit extracts a possible operation target for the operation content specified by the operation specifying unit as an operation target candidate.

  Thereby, with respect to an ambiguous word, the operation target candidate according to the specified operation content can be efficiently extracted, and the operation target can be specified accurately.

  In another embodiment, when there are a plurality of operation objects and operation contents specified by the operation specifying means, a notification means for notifying the plurality of operation objects and operation contents, and among the plurality of operation objects and operation contents Selection means for selecting one each, and the operation specifying means sets the operation target and operation content selected by the selection means as the operation target and operation content finally specified.

  Accordingly, the user can be informed about a plurality of operation objects and a desired one can be selected, so that the operation objects can be accurately identified.

  Still another embodiment includes a device state detection unit that detects a state of a device to be operated, and when there are a plurality of operation targets and operation contents specified by the operation specifying unit, the operation specifying unit includes: One of each of the plurality of operation objects and operation contents is specified based on a detection result by the device state detection means.

  Thereby, even when there are a plurality of specified operation targets and operation details, the most appropriate operation target and operation details can be specified from the state of the device to be operated.

  Still another embodiment includes an operation history storage unit that stores a history of operation targets and operation details, and when there are a plurality of operation targets and operation details specified by the operation specifying unit, the operation specifying unit includes: One of each of the plurality of operation objects and operation contents is specified based on a history stored in the operation history storage means.

  Thereby, even when there are a plurality of specified operation objects and operation contents, the most appropriate operation object and operation contents can be specified from the operation history so far.

  In addition, according to the present invention, there is provided a speech recognition method for determining an operation content of an operation target based on a recognition result for an input voice, wherein the recognition result for the voice is determined based on a predetermined operation target and operation content. The operation step and the operation content can be specified by the classification step, and the operation content can be specified in the specification step, but when the operation target cannot be specified, the identification result is obtained from the recognition result of the voice. An ambiguous word detection step for detecting an ambiguous word that cannot be classified in the step; a candidate extraction step for extracting an operation target that may be specified by the ambiguous word detected in the ambiguous word detection step; The ambiguous word detected in the detection step is replaced with the candidate extracted in the candidate extraction step, and the recognition result after the replacement is determined as a predetermined operation pair. And by classifying the type of operation contents, the method characterized by comprising a second specifying step of specifying the operation target and operation content is provided.

  According to this speech recognition method, as described with reference to the speech recognition apparatus of the present invention, it is possible to cope with a case where an ambiguous word that does not specify the operation target is included in the user's utterance. It is possible to estimate an appropriate operation by extracting candidates to be performed. Therefore, according to this speech recognition method, it is possible to cope with the case where an ambiguous word whose operation target cannot be specified is included in the user's utterance, and it is appropriate to extract candidates for specifying the operation target. Can be estimated.

  Furthermore, according to the present invention, there is provided a speech recognition program for causing a computer to execute a process of determining an operation content of an operation target based on a recognition result for input speech, wherein the recognition result for the speech is determined in advance. A specific process for identifying the operation target and the operation content by classifying the operation target and the operation content, and a detection process for detecting an ambiguous word that cannot be classified by the specific process from the recognition result of the voice. A candidate extraction process for extracting an operation target that may be identified by an ambiguous word detected in the detection process as an operation target candidate, and the operation content can be identified in the identification process. If the computer cannot perform the specific processing by replacing the ambiguous word detected by the detection processing with the candidate extracted by the extraction processing, Speech recognition program for causing a is provided.

  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.

  As shown in FIG. 1, the speech recognition apparatus of the present invention comprises 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 operated 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 operation 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 voice recognition 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. Further, the voice interaction unit 1 determines a scenario using the scenario database 18 based on the operation candidate specified from the recognition result of the utterance, and responds to the driver, controls the device, and the like, And a speech synthesis unit 14 that synthesizes a response by speech output to the driver using the phoneme model 21. The “operation candidate” corresponds to an operation target or operation content candidate specified based on the recognition result of the utterance.

  More specifically, the syntax analysis unit 12 includes, as its functions, an operation specifying unit 31 that performs a specifying process (a “syntax analyzing process” described later) for specifying an operation target and an operation content from the recognized text, and an operation specifying unit. The ambiguous word detection means 32 that detects ambiguous words that cannot be classified by the reference numeral 31 and the operation target that may be identified by the ambiguous words detected by the ambiguous word detection means 32 are processed using the conversion candidate database 34. Candidate extraction means 33 for extracting as candidates.

  Further, the operation specifying unit 31 displays a plurality of operation candidates included in the operation candidate group specified by the specifying process via the touch panel 24 so as to prompt the driver to select, and via the speaker 4, A voice guide that prompts the driver to select is output. The operation specifying unit 31 specifies a final operation candidate based on a touch operation on the touch panel 24 from the driver. The scenario control unit 13 and the speech synthesis unit 14 constitute notification means of the present invention. Further, the scenario control unit 13 constitutes a selection unit of the present invention.

  Note that the acoustic model 15, the language model 16, the syntax model 17, the scenario database 18, the phoneme model 19, the conversion candidate database 34, and the recording medium (database) such as a CD-ROM, DVD, and HDD in which data is recorded. ).

  Further, since the operation history storage unit 35 indicated by a broken line in FIG. 1 has a configuration provided only in a third embodiment to be described later, description thereof is omitted here.

  The voice recognition unit 11 performs frequency analysis on the waveform data indicating the voice of the utterance input to the microphone 2 and extracts a feature vector. Then, the speech recognition unit 11 executes “speech recognition processing” that recognizes the input speech based on the extracted feature vector and outputs the recognized speech as a text represented by a word string. This speech recognition process is executed by comprehensively determining the acoustic features and linguistic features of the input speech using a probabilistic method as described below.

  That is, the speech recognition 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 speech recognition 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 speech recognition unit 11 determines the certainty of speech recognition for all the determined texts based on the acoustic score and language score of the text (hereinafter, this certainty is referred to as “speech recognition score” as appropriate). calculate. Then, the speech recognition unit 11 outputs the text expressed by a word string whose speech recognition score satisfies a predetermined condition as 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 speech recognition 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 speech recognition 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. The parsing unit 12 then classifies the classified text (Categorized Text) that satisfies a predetermined condition of the parsing score as an operation candidate group that is specified based on the recognition result of the input utterance, together with the parsing score. Output. The “class” corresponds to a classification according to a category representing an operation target or operation content, such as the domain described above. For example, if the recognized text is “setting change”, “setting change”, “change setting”, or “setting change”, the classified text becomes {Setup}.

  At this time, the syntax analysis unit 12 detects an ambiguous word included in the recognized text when the operation content can be specified from the text recognized by the speech recognition unit 11 by the syntax analysis process but the operation target cannot be specified. The ambiguous word is replaced with an operation target candidate using the conversion candidate database 34, and the syntax analysis process is performed again on the replaced text.

  In addition, the scenario control unit 13 outputs a response to the driver using data recorded in the scenario database 18 based on the specified operation candidate and the state of the vehicle 10 acquired from the vehicle state detection unit 3. And determine device control scenarios. A plurality of scenarios for response output and device control are recorded in the scenario database 18 together with operation candidates and vehicle condition conditions. 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) or a value that is sequentially set so that the processing time of the speech recognition processing for the input utterance is within a predetermined time 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 voice recognition 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 operation target is the audio 6a. {Climate} indicates that the operation target is the air conditioner 6c. {Passenger Climate} indicates that the operation target is the air conditioner 6c in the passenger seat. {POI} indicates that the operation target is the POI search function of the navigation system 6b. {Navigation} indicates that the operation target is a function such as route guidance or map operation of the navigation system 6b. {Clock} indicates that the operation target is a clock function. {Help} indicates that the operation target is a help function for knowing how to operate the devices 6a to 6c and the speech recognition apparatus. {Ambiguous} indicates that the operation target is unknown.

  In addition, in the conversion candidate database 34, one or a plurality of operation target candidates corresponding to ambiguous words and classified texts are recorded in advance as data. Here, the ambiguous word is a general instruction word such as “this”, “it”, “that”. For example, as shown in FIG. 3, when the ambiguous word is “it” and the classified text is {Ambiguous_Setting_Up}, the operation target candidates are “song”, “station”, “disk”, “volume”, “ The temperature, fan, scroll, and scale are seven. Further, for example, when the ambiguous word is “it” and the classified text is {Ambiguous_Setting_Max}, the conversion candidates are “volume”, “temperature”, “fan”, and “scale”. “Volume” indicates that the operation target is the volume of the audio 6a. “Fan” indicates that the operation target is the speed (air volume) of the fan of the air conditioner 6c.

  Next, the operation (voice dialogue processing) of the voice recognition device of this embodiment will be described. As shown in FIG. 4, 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 voice recognition 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 spoken dialogue unit 1 determines the text expressed by the word string based on the language score of the text using the data of the entire language model 16 from the determined pronunciation data. At this time, when a plurality of pronunciation data are determined, text is determined for each 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 illustrated in FIG. 5, when the input utterance is “Set the station ninety nine point three FM.”, The text corresponding to the pronunciation data determined from the 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 voice interaction unit 1 calculates the weighted sum of the acoustic score and the language score for all the determined texts as the certainty of voice recognition (voice recognition score). As the weighting factor, for example, a value predetermined experimentally is used.

  Next, the voice interaction unit 1 determines and outputs the text represented by the word string whose calculated voice recognition score satisfies the predetermined condition as the recognized text. The predetermined condition is determined in advance, for example, as a text having the highest speech recognition score, a text having a speech recognition score from a higher rank to a predetermined rank, or a text having a speech recognition 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. Specifically, the voice interaction unit 1 uses the syntax model 17 to determine the classified text from the recognized text.

  First, the voice interaction unit 1 calculates the likelihood of each domain in one word for each word included in the recognized text, using data of the entire syntax model 17. Next, the voice interaction unit 1 determines a domain in one word based on the likelihood. Next, the voice interaction unit 1 uses the data of the portion classified into the domain of the determined type in the syntax model 17, and the likelihood (words) of each class set (classified text) in one word Score). Then, the voice interaction unit 1 determines the classified text in one word based on the word score.

  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.

  Here, the process for determining the text classified by using the syntax model 17 will be described using the example shown in FIG. In the example of FIG. 6, 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.

  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. For example, the parsing score for {Climate_ACOnOff_On} is i1 + j1. For example, the parsing score for {Climate_Defrost_Front} is i5 + j4. Then, based on the calculated parsing score, a class set (classified text) for the entire text is determined. As a result, text classified into {Climate_Defrost_Front} {Climate_Fan-Vent_Floor} {Climate_ACOnOff_On} is determined from the recognized text.

  Next, the voice interaction unit 1 uses the classified text (Categorized Text) whose calculated parsing score satisfies a predetermined condition as an operation candidate specified based on the recognition result of the input utterance. And the certainty (syntactic analysis score) of the operation candidate. The predetermined condition is determined in advance as, for example, text having the highest parsing score, text having a parsing score from the top to a predetermined rank, text having a parsing score of a predetermined value or more, and the like. For example, when the utterance “AC on floor to defrost” is input as described above, {Climate_Defrost_Front} is output as an operation candidate together with its parsing score.

  Next, in STEP 4, the voice interaction unit 1 determines whether or not the operation content is identified from the operation candidates. Specifically, the voice interaction unit 1 determines whether or not the class classified text obtained as a result of the syntax analysis processing in STEP 3 includes a class other than the type of domain at the head portion. When the determination result of STEP 4 is YES (when the operation content is not specified), the process proceeds to STEP 10 and the result of the syntax analysis process of STEP 3 is output as it is as an operation candidate group.

  If the determination result in STEP 4 is NO (operation content is specified), the process proceeds to STEP 5 and the voice interaction unit 1 determines whether or not the operation target is specified from the operation candidates. Specifically, the voice interaction unit 1 determines whether or not the domain type of the classified text obtained as a result of the syntax analysis process in STEP 3 is {Ambiguous}. If the determination result in STEP 5 is NO (the domain type is not {Ambiguous}, that is, the operation target is specified), the process proceeds to STEP 10 and the result of the syntax analysis process in STEP 3 is output as it is as an operation candidate group. .

  If the determination result in STEP 5 is YES (the domain type is {Ambiguous}, that is, the operation target is not specified), the process proceeds to STEP 6 and the voice interaction unit 1 includes an ambiguous word in the text recognized in STEP 2. It is determined whether or not. Specifically, it is determined whether or not general instructions such as “this”, “it”, and “that” are included. If the determination result in STEP 6 is NO (no ambiguous word is included), the process proceeds to STEP 10 and the result of the syntax analysis process in STEP 3 is output as it is as an operation candidate group. By these determination processes in STEPs 3 to 5, ambiguous words included in the recognized text are detected.

  If the determination result in STEP 6 is YES (an ambiguous word is included), the process proceeds to STEP 7 and the speech dialogue unit 1 replaces the ambiguous word included in the recognized text with a direct word indicating the operation target candidate. Execute the process. First, the voice interaction unit 1 reads an operation target candidate from the conversion candidate database 34 based on the ambiguous word and the classified text. Then, the voice interaction unit 1 replaces the ambiguous word portion of the recognized text with the read operation target candidate.

  Next, in STEP 8, the voice interaction unit 1 executes a parsing process for each replaced text in the same manner as in STEP 3. As a result, for each text after replacement, the classified text is determined together with the parsing score of the classified text. This parsing score indicates the validity of the text after replacement.

  Next, in STEP 9, the voice interaction unit 1 is input with the classified text (Categorized Text) whose parsing score satisfies a predetermined condition among the classified text obtained in STEP 8. The operation candidate specified based on the recognition result of the utterance is output together with the certainty of the operation candidate (syntactic analysis score). The predetermined condition is determined in advance as, for example, text having the highest parsing score, text having a parsing score from the top to a predetermined rank, text having a parsing score of a predetermined value or more, and the like. Thereby, operation candidates corresponding to the user's speech can be included in the operation candidate group with high probability.

  Here, the processing of STEP 4 to 9 will be described using the example shown in FIG. The example of FIG. 7 is a case where the utterance “make it stronger” is input from the driver in STEP 1 as shown in FIG. In this case, as shown in FIG. 7B, the text recognized by the speech recognition process in STEP 2 becomes “strengthen”, and as shown in FIG. 7C, as a result of the syntax analysis process in STEP 3 The classified text {Ambiguous_Setting_Up} is obtained.

  At this time, since {Ambiguous_Setting_Up} includes a class {Setting_Up} other than the type of the domain at the head portion and the control content is specified, the determination result in STEP 4 is NO and the process proceeds to STEP 5. Next, since the domain type of {Ambiguous_Setting_Up} is {Ambiguous} and the operation target is not specified, the determination result in STEP 5 is YES, and the process proceeds to STEP 6. Since the ambiguous word “it” is included in the text “make it stronger”, the determination result in STEP 6 is NO. Thereby, the word “it” included in the recognized text is detected as an ambiguous word.

Next, in STEP 7, based on the text {Ambiguous_Setting_Up} classified as the ambiguous word “it” from the conversion candidate database 34 as shown in FIG. 3, “song”, “station”, “disk”, “volume” are converted as conversion candidates. “Temperature” “Fan” “Scroll” “Scale” is read. Then, the ambiguous words contained in the text “Strengthen” are replaced with these conversion candidates. As a result of this replacement process, as shown in FIG. 7 (d), as the replaced text A ₁ to A _7, "Song strongly to""Volume strongly to""Fan strongly to""strongly scale Or the like.

Next, in STEP 8, the post-replacement text _A 1 to A _7, parsing process is executed, as shown in the table of FIG. 7 (d), the as text _B 1 .about.B ₇ which is classification, {Audio_CD_NextTrack} Are obtained together with the parsing scores Sc (B ₁ ) to Sc (B ₇ ) of the texts B _{1 to} B ₇ .

Next, in STEP 9, based on the parsing score _{Sc (B 1) ~Sc (B} 7), the validity of the replaced text _A 1 to A ₇ are evaluated, the candidate operation is outputted is specified. For example, texts such as “Strengthen the song” and “Strengthen the scale” are unnatural as word connections, so the parsing score is low as shown in FIG. On the other hand, for example, a text such as “enhance volume” and “enhance fan” has a relatively valid word connection, and therefore has a high parsing score as shown in FIG. Therefore, for example, when the predetermined threshold is 0.2, among the results of the syntax analysis process, those results having a low syntax analysis score are not included in the operation candidates. As a result, the first operation candidate L ₁ {Audio_Volume_Up} and the second operation candidate L ₂ {Climate_Fan_Up} are output as operation candidate groups as shown in FIG.

  Returning to FIG. 4, next, in STEP 11, 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 vehicle 10) detected by the vehicle state detection unit 3. The detected value of the driver's state, etc.) is acquired.

  Next, in STEP 11, the voice interaction unit 1 executes a process of specifying a final operation candidate from the operation candidate group output in STEP 9 and 10. Specifically, the voice interaction unit 1 displays the one or a plurality of operation candidates obtained in STEPs 9 and 10 on the touch panel 24 so as to prompt the driver to select. At the same time, the voice interaction unit 1 synthesizes a voice guide that prompts the driver to select an operation candidate by the voice synthesizer 14 and outputs it from the speaker 4.

For example, in the example of FIG. 7 described above, as shown in FIG. 8, the display “Increase volume” corresponding to the first operation candidate L ₁ and “Increase the air volume of the air conditioner” corresponding to the second operation candidate L _2. A sound display is displayed on the touch panel 24, and a sound that prompts the driver to confirm whether or not the _first and second operation candidates L ₁ are correct is output from the speaker. Thus, the first candidate operation L ₁ and second candidate operation L ₂ is displayed simultaneously or sequentially the touch panel 24. For example, as shown in FIG. 8 (a), first an operation candidate L ₁ "Volume Up" and a second candidate operation L ₂ "increase the air volume of the air conditioner" is displayed on the touch panel 24 at the same time. Also, as shown in FIG. 8B, first, “Increase volume” as the first operation candidate L ₁ is displayed on the touch panel 24, and FIG. 8 ( as shown in c) is the second candidate operation L ₂ "increase the air volume of the air conditioner" is displayed on the touch panel 24.

  At this time, the parsing process is performed on the text in which the ambiguous word is replaced in STEPs 4 to 9, and the operation candidates are identified by more clearly grasping the meaning of the utterance. Therefore, the operation candidates displayed on the touch panel 24 are displayed on the screen. The operation candidates corresponding to the driver's utterance can be included with high probability, and the driver can select the operation candidate according to the intention of the utterance.

Next, the voice interaction unit 1 identifies a final operation candidate based on a touch operation on the touch panel 24 from the driver. For example, the driver performs touch input for selecting one of the plurality of operation candidates L ₁ and L ₂ displayed on the touch panel 24. Then, the voice interaction unit 1 specifies the operation candidate selected by this touch input as the final operation candidate. Thereby, the operation candidate that matches the driver's utterance is specified as the final operation candidate.

  Next, in STEP 13, the voice interaction unit 1 uses the scenario database 18 based on the operation candidate specified in STEP 12 and the state of the vehicle 10 detected in STEP 11 to output a response to the driver and control the equipment. Determine the scenario for.

  First, the voice interaction unit 1 acquires information for controlling a target from the identified operation candidates. As shown in FIG. 9, 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 interaction unit 1 inputs a value into a slot of the corresponding form based on the operation candidate specified from the recognition result of each utterance in the dialogue with the driver. 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. The certainty of the form is calculated based on, for example, the certainty of the operation candidate specified from the recognition result of each utterance and the filling degree of the slot of each form. For example, as shown in FIG. 10, when an utterance “Guide to Chitose Airport with the shortest route” 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. Then, based on the selected form, the scenario is determined using the data stored in the scenario database 18. As shown in FIG. 10, the scenario database 18 stores, for example, response sentences to be output to the driver, classified by 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. 10, 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 14, 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 STEP14 is NO, the process proceeds to STEP15, where the voice interaction unit 1 synthesizes a voice according to the contents of the determined response sentence and the conditions for outputting the response sentence. In STEP 16, 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 13, the voice interaction unit 1 executes processing for the second utterance in the same manner as 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 14, the voice interaction unit 1 determines whether or not the dialogue with the driver has ended. When the determination result in STEP 14 is NO, the voice interaction unit 1 executes the processing in STEPs 15 to 16 as in the first utterance.

  Thereafter, the processing similar to STEP 1 to STEP 16 for the second utterance described above is repeated until the determination result in STEP 14 is YES.

  If the determination result in STEP 14 is YES, the process proceeds to STEP 17, where 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, in STEP 18, a response sentence is output from the speaker 4. Next, in STEP 19, the voice interaction unit 1 controls the device based on the determined scenario, and ends the voice interaction process.

  With the above processing, the content indicated by the ambiguous word included in the utterance is appropriately estimated and the driver's intention is more clearly grasped, so that the device is controlled through efficient dialogue.

In the present embodiment, in the process of specifying the final operation candidate in STEP 12, the operation candidate is selected and the operation candidate to be displayed is changed by a touch input from the driver to the touch panel 24. The operation candidate can be selected by, for example, voice input by the driver (an utterance for selecting any operation candidate or an utterance of “next”). Further, an input interface (keyboard, button, switch, multi-jog dial, etc.) is provided, and selection of operation candidates and the like can be performed by inputting to the input interface. In these cases, operation candidates may be displayed on the display 5 other than the touch panel 24.
[Second Embodiment]
Next, the speech recognition apparatus according to the second embodiment of the present invention will be described. Note that the present embodiment is different from the first embodiment only in the process of specifying the final operation candidate in STEP 12 of the voice interaction process. 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 voice recognition device according to the present embodiment, the operation specifying unit 31 is finally determined based on the states of the devices 6a to 6c mounted on the vehicle 10 detected by the vehicle state detection unit 3 from the specified operation candidate group. Specific operation candidates are identified. Other configurations are the same as those of the first embodiment. The vehicle state detection unit 3 corresponds to the device state detection unit of the present invention.

  In the operation of the voice recognition device of the present embodiment (voice dialogue processing), in STEP12, first, the voice dialogue unit 1 determines the STEPs 9 and 10 among the plurality of devices 6a to 6c from the detection result of the vehicle state detection unit 3. The state of the device related to the operation candidate group obtained from the process is acquired. Then, based on the state of the device, an operation candidate that is more likely to correspond to the driver's intention is determined from the operation candidate group.

Here, the case where the first and second operation candidates L ₁ and L ₂ are obtained as shown in FIG. 7 will be described as an example. At this time, the devices related to the operation candidate group are the audio 6a and the air conditioner 6c, and the volume of the audio 6a and the air volume of the air conditioner 6c are acquired. The volume of the audio 6a is the maximum value, the air volume of the air conditioner 6c is 3, and the maximum value of the air volume is 5. In this case, since the volume cannot be increased from the current state, it is estimated that the operation is performed by the air conditioner 6c. Therefore, {Climate_Fan_Up} is determined as the final operation candidate. 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, the content indicated by the ambiguous word included in the utterance is appropriately estimated and the driver's intention is more clearly grasped, which is efficient. The device is controlled through 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 specifying the final operation candidate in STEP 12 of the voice interaction process. 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 voice recognition apparatus according to the present embodiment includes an operation history storage unit (operation history storage unit) 35 that stores a history of operation targets and operation contents (operation history). The operation history storage unit 35 stores the operation details of the devices 6a to 6c by the driver together with the date and time of the operation as data. Furthermore, based on these data, the frequency of use of the devices 6a to 6c by the driver, the number of times of use, etc. are grasped.

  Then, the operation specifying unit 31 specifies a final operation candidate based on the operation history of the devices 6a to 6c by the driver stored in the operation history storage unit 35 from the specified operation candidate group. Specifically, for example, an operation candidate is specified based on the operation performed immediately before the driver (the previous operation), the operation performed immediately before the operation (the operation performed two times before), or the like. . Other configurations are the same as those of the first embodiment.

  In the operation (voice dialogue processing) of the voice recognition device of the present embodiment, in STEP 12, first, the voice dialogue unit 1 reads the driver's operation history from the operation history storage unit 34. Then, based on the operation history, an operation candidate that is more likely to correspond to the driver's intention is identified from one or a plurality of operation candidates obtained from STEPs 9 and 10.

Here, the case where the first and second operation candidates L ₁ and L ₂ are obtained as shown in FIG. 7 will be described as an example. At this time, from the operation history storage unit 34, the previous operation may be an operation for lowering the temperature setting of the air conditioner 6c, and the second operation may be an operation for turning on the air conditioner 6c. Is read. And since the operation with respect to the air conditioner 6c is continuing, it is presumed that the operation intended for the utterance this time is also likely to be operated with respect to the air conditioner 6c. Therefore, {Climate_Fan_Up} is specified as the final operation candidate. 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, the content indicated by the ambiguous word included in the utterance is appropriately estimated and the driver's intention is more clearly grasped, which is efficient. The device is controlled through dialogue.

  In addition, about the process which specifies the final operation candidate in STEP12 of a voice interaction process, it is also possible to use combining two or all three of the processes in the first to third embodiments. For example, from the operation candidate group obtained from STEPs 9 and 10, one or a plurality of operation candidates that are highly likely are determined based on the state of the device and the operation history. Further, the determined operation candidate is displayed via the touch panel 24, and the final operation candidate is specified based on the touch input from the driver.

  In the first to third embodiments, the scenario control unit 13 determines the scenario based on the operation candidate specified from the utterance recognition result and the vehicle state detected by the vehicle state detection unit 3. The scenario control unit 13 may determine the control process only from the operation candidates specified from the utterance 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. Explanatory drawing which shows the structure of the conversion candidate database in the speech recognition apparatus of FIG. The flowchart which shows the whole operation | movement (voice dialogue process) of the speech recognition apparatus of FIG. Explanatory drawing which shows the speech recognition process using the language model in the speech dialogue process of FIG. FIG. 5 is an explanatory diagram illustrating a syntax analysis process using a syntax model in the voice interaction process of FIG. 4. Explanatory drawing which shows the process which replaces the ambiguous word in the voice dialogue process of FIG. 4, and specifies an operation candidate. Explanatory drawing which shows the process which specifies the final operation candidate 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 interaction process of FIG. Explanatory drawing which shows the process which determines the scenario in the voice dialogue process of FIG.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Voice dialogue unit, 2 ... Microphone, 3 ... Vehicle state detection part, 4 ... Speaker, 5 ... Display, 6a-6c ... Equipment, 10 ... Vehicle, 11 ... Voice recognition part, 12 ... Syntax analysis part, 13 ... Scenario Control unit, 14 ... speech synthesis unit, 15 ... acoustic model, 16 ... language model, 17 ... syntax model, 18 ... scenario database, 19 ... phoneme model, 24 ... touch panel, 31 ... operation specifying means, 32 ... ambiguity word detecting means 33 ... Candidate extraction means, 34 ... Conversion candidate database, 35 ... Operation history storage unit.

Claims (7)

In the speech recognition apparatus that determines the operation content of the operation target based on the recognition result of the input speech,
An operation specifying means for performing a specifying process for specifying the operation target and the operation content by classifying the recognition result of the voice into a predetermined type of operation target and operation content;
An ambiguous word detecting means for detecting an ambiguous word that cannot be classified by the operation specifying means from the recognition result of the voice;
Candidate extraction means for extracting an operation target that may be identified by the ambiguous word detected by the ambiguous word detection means as an operation target candidate,
If the operation content can be specified and the operation target cannot be specified, the operation specifying unit replaces the ambiguous word detected by the ambiguous word detecting unit with the candidate extracted by the candidate extracting unit, and A speech recognition apparatus characterized by performing the specifying process by specifying means. The speech recognition apparatus according to claim 1,
The speech recognition apparatus, wherein the candidate extracting unit extracts an operation target that has a possibility of the operation content specified by the operation specifying unit as an operation target candidate. The speech recognition apparatus according to claim 1 or 2,
When there are a plurality of operation targets and operation contents specified by the operation specifying means, a notification means for notifying the plurality of operation objects and operation contents and one of each of the plurality of operation objects and operation contents are selected. Selecting means,
The speech recognition apparatus characterized in that the operation specifying means uses the operation target and operation content selected by the selection means as the final operation target and operation content. The speech recognition apparatus according to claim 1 or 2,
Provided with device status detection means for detecting the status of the device to be operated,
When there are a plurality of operation objects and operation contents specified by the operation specifying means, the operation specifying means is configured to select one of the plurality of operation objects and operation contents based on a detection result by the device state detection means. A voice recognition device characterized by identifying one. The speech recognition apparatus according to claim 1 or 2,
An operation history storage means for storing a history of the operation target and operation content is provided,
When there are a plurality of operation targets and operation contents specified by the operation specifying means, the operation specifying means, based on the history stored in the operation history storage means, the plurality of operation targets and operation contents. A speech recognition apparatus characterized by identifying one of each. A speech recognition method for determining an operation content to be operated based on a recognition result of input speech,
A step of identifying the operation target and the operation content by classifying the recognition results for the voice into predetermined types of operation target and operation content; and
The operation content can be specified in the specifying step, but when the operation target cannot be specified, the ambiguous word detecting step of detecting the ambiguous word that cannot be classified in the specifying step from the recognition result of the voice;
A candidate extraction step of extracting an operation target that may be specified by the ambiguous word detected in the ambiguous word detection step as an operation target candidate;
By replacing the ambiguous word detected in the detection step with the candidate extracted in the candidate extraction step, and classifying the recognition result after the replacement into a predetermined type of operation target and operation content. And a second specifying step for specifying the operation content. A speech recognition program for causing a computer to execute a process of determining an operation content of an operation target based on a recognition result of input speech,
A specific process for identifying the operation target and the operation content by classifying the recognition result of the voice into a predetermined type of operation target and operation content;
From the recognition result for the voice, a detection process for detecting an ambiguous word that cannot be classified by the specific process;
A candidate extraction process that extracts an operation target that may be identified by an ambiguous word detected in the detection process as an operation target candidate,
The function of performing the specifying process by replacing the ambiguous word detected by the detection process with the candidate extracted by the extraction process when the operation content can be specified by the specifying process and the operation target cannot be specified A speech recognition program which is executed by a computer.

Images