JP2016180916A - Voice recognition system, voice recognition method, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a voice recognition system capable of reducing the number of false recognition and improving utilization efficiency of a system.SOLUTION: A client device 10 includes: a receiving unit 15 for receiving voice recognition results from a voice recognition server device group 20 selected based on a voice collection condition; a correct answer candidate extracting unit for extracting a voice recognition result of a last signal from a repeated signal group as a correct answer candidate; a transmission unit 14 for transmitting, to a control unit, a pair of the correct answer candidate and a re-learning signal group; and a transmission destination changing unit 18 for changing a relation between a voice recognition server device functioning as a transmission destination of an acoustic signal and the voice collection condition on the basis of a transmission destination information. A control unit 30 includes: a voice recognition result receiving part for receiving voice recognition results for the re-learning signal group from all the voice recognition server devices 21; a transmission destination information updating part for updating transmission destination information on the basis of a similarity between each of the voice recognition results received from all the voice recognition server devices and the correct answer candidate; and a transmission destination information transmitting part for transmitting the updated transmission destination information to the client device.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a voice recognition system, a voice recognition method, and a program including a client device, a plurality of voice recognition server devices, and a management unit.

  2. Description of the Related Art Conventionally, there is a server / client type speech recognition system that performs speech recognition on a speech section signal detected by a client device and returns the result to the client device (for example, Patent Document 1). By arranging the voice recognition server device on a network accessible from any client device, a number of client devices can enjoy services using the voice recognition system.

JP 2005-331616 A

  At this time, if the recognition performance of the voice recognition server device is not sufficient, the client device needs to access the voice recognition server device many times until a correct recognition result is obtained. This multiple access increases the load on the voice recognition server device and reduces the system utilization efficiency. In order to improve the utilization efficiency of the system, it is necessary to reduce the number of erroneous recognitions.

  Therefore, an object of the present invention is to provide a speech recognition system that can reduce the number of erroneous recognitions and improve the utilization efficiency of the system.

  The speech recognition system of the present invention includes a client device, a plurality of speech recognition server devices, and a management unit. The client device includes a receiving unit, a rephrase determining unit, a transmitting unit, and a transmission destination changing unit.

  The receiving unit receives a speech recognition result for the acoustic signal input to the client device from the speech recognition server device selected based on the sound collection condition. The rephrasing determination unit acquires a repeated signal group that is a signal group obtained by observing a plurality of repetitions of the utterance indicating the same content by the user, and extracts the speech recognition result of the last signal from the repeated signal group as a correct candidate To do. The transmission unit sets all repetitive signal groups as relearning signal groups, and transmits a set of correct answer candidates and relearning signal groups to the management unit. The transmission destination changing unit includes a voice recognition server device serving as a transmission destination of the acoustic signal and a sound collection condition based on transmission destination information that is information regarding a relationship between the voice recognition server device serving as the transmission destination of the acoustic signal and the sound collection condition. Change the relationship.

  The management unit includes a voice recognition result reception unit, a transmission destination information update unit, and a transmission destination information transmission unit.

  The speech recognition result receiving unit receives speech recognition results for the relearning signal group from all speech recognition server devices. The transmission destination information update unit updates the transmission destination information based on the similarity between each voice recognition result received from all the voice recognition server devices and the correct answer candidates. The transmission destination information transmission unit transmits the updated transmission destination information to the client device.

  According to the speech recognition system of the present invention, the number of erroneous recognitions can be reduced, and the utilization efficiency of the system can be improved.

1 is a block diagram illustrating a configuration of a voice recognition system according to Embodiment 1. FIG. The block diagram which shows the structure of the rephrase determination part of the speech recognition system of Example 1. FIG. FIG. 2 is a block diagram illustrating a configuration of a management unit of the voice recognition system according to the first embodiment. FIG. 3 is a sequence diagram illustrating a voice recognition operation of the voice recognition system according to the first embodiment. FIG. 3 is a sequence diagram illustrating an information update operation of the voice recognition system according to the first embodiment. 3 is a flowchart illustrating the operation of a rephrase determining unit of the voice recognition system according to the first embodiment. The figure which illustrates rephrasing determination operation | movement of the speech recognition system of Example 1. FIG. The figure which illustrates the transmission destination information update operation | movement of the speech recognition system of Example 1. FIG. FIG. 3 is a block diagram illustrating a configuration of a voice recognition system according to a second embodiment. FIG. 6 is a block diagram illustrating a configuration of a management unit of the voice recognition system according to the second embodiment. FIG. 9 is a sequence diagram illustrating an information update operation of the voice recognition system according to the second embodiment. FIG. 6 is a block diagram illustrating a configuration of a voice recognition system according to a third embodiment. FIG. 9 is a block diagram illustrating a configuration of a management unit of the voice recognition system according to the third embodiment. FIG. 10 is a sequence diagram illustrating an information update operation of the voice recognition system according to the third embodiment.

  Hereinafter, embodiments of the present invention will be described in detail. In addition, the same number is attached | subjected to the structure part which has the same function, and duplication description is abbreviate | omitted.

  In the following description, a signal uttered as a speech recognition target is referred to as a sound signal, and a signal collected in a state where a sound signal and a background noise signal other than the sound signal are mixed is referred to as an acoustic signal.

  Hereinafter, the configuration of the speech recognition system according to the present embodiment will be described with reference to FIGS. 1, 2, and 3. FIG. 1 is a block diagram showing the configuration of the speech recognition system 1 of the present embodiment. FIG. 2 is a block diagram illustrating a configuration of the rephrase determining unit 17 of the speech recognition system 1 according to the present embodiment. FIG. 3 is a block diagram illustrating a configuration of the management unit 30 of the voice recognition system 1 according to the present embodiment.

  As shown in FIG. 1, the speech recognition system 1 of the present embodiment includes a client device 10 and a plurality of speech recognition server devices 21-1, ..., 21-n, ..., 21-N (N is N ≧ 2). An integer to be satisfied, n is an integer satisfying 1 ≦ n ≦ N), and the management unit 30. Although only one client device 10 is shown in FIG. 1, a plurality of client devices 10 may exist. When the voice recognition server devices 21-1,..., 21-n,..., 21-N are collectively called, they are referred to as a voice recognition server device group 20. It is assumed that the client device 10 and the voice recognition server device group 20 are connected to be communicable wirelessly or by wire via a network. The management unit 30 may be configured as a single hardware (device). When the management unit 30 is configured as a single piece of hardware (device), it may be called the management device 30. When the management unit 30 is configured as a single piece of hardware (device), the client device 10, the voice recognition server device group 20, and the management unit 30 (management device 30) are connected to be communicable wirelessly or wired via a network. It shall be. Further, the management unit 30 may be a configuration requirement in the client device 10 or a configuration requirement in any voice recognition server device in the voice recognition server device group 20.

  The voice recognition server devices 21-1,..., 21-n,..., 21-N are in charge of the sound recognition processing of the sound signals based on the sound collection conditions (details will be described later) of the sound signals input to the client device 10. It is assumed that whether or not to perform is set in advance and acoustic models having different characteristics are stored. As the characteristics of the acoustic model, for example, noise characteristics can be considered. The client device 10 includes a sound collection condition extraction unit 11, a threshold storage unit 111, a selection unit 12, a transmission destination storage unit 121, a signal processing unit 13, a transmission unit 14, a reception unit 15, and a presentation. A re-statement determination unit 17; a re-statement information storage unit 171; and a transmission destination change unit 18. As shown in FIG. 2, the rephrasing determination unit 17 of the present embodiment includes a reaction time measurement unit 17A, a reliability acquisition unit 17B, a similarity calculation unit 17C, and a determination unit 17D. As shown in FIG. 3, the management unit 30 of the present embodiment includes a correct candidate receiving unit 30A, a relearning signal group transmitting unit 30B, a speech recognition result receiving unit 30C, a transmission destination information updating unit 30D, and transmission destination information. A transmission unit 30E and a correct candidate storage unit 30F are included.

  Hereinafter, the speech recognition operation of this system will be described with reference to FIG. FIG. 4 is a sequence diagram showing the voice recognition operation of the voice recognition system 1 of the present embodiment. First, the sound collection condition extraction unit 11 extracts a sound collection condition of the input acoustic signal (S11). The selection unit 12 selects a voice recognition server device (for example, the voice recognition server device 21-1) as a transmission destination of the corresponding voice signal based on the extracted sound collection condition (S12). It is assumed that the relationship between the sound collection condition and the voice recognition server device as the transmission destination is stored in advance in the transmission destination storage unit 121 as transmission destination information.

<Sound collection conditions>
The sound collection conditions include, for example, a feature amount relating to the S / N ratio, which is a ratio of the size of the audio signal and the background noise signal, a feature amount relating to the distortion of the acoustic signal, a feature amount relating to the spectrum shape of the background noise signal, and a background noise signal. It is possible to set a condition based on a threshold value for at least one of the feature amounts related to the size of the feature amount. It is assumed that the threshold value is stored in advance in the threshold value storage unit 111.

  The background noise signal is a signal observed by a microphone for a certain time immediately before the voiced sound or the target sound is input. The magnitude of the background noise signal is an average value of the power spectrum of the background noise signal over a certain period of time. The spectrum shape of the background noise signal is a component of each band in the spectrum of the background noise signal and its time change. The S / N ratio between the audio signal and the background noise signal is the ratio of the size of the audio signal in the acoustic signal being input to the uttered speech (target sound) and the size of the background noise signal. As the audio signal, a power spectrum obtained by subtracting the average value of the power spectrum of the background noise signal for a certain time from the power spectrum of the acoustic signal for a certain time during the input of the uttered voice (target sound) can be used. The magnitude of the audio signal is an average value of the power spectrum of the audio signal for a certain time during the input of the uttered voice (target sound).

  The distortion of the acoustic signal means that the microphone element, the microphone amplifier circuit, and the A / D conversion are clipping because the input of the acoustic signal is too large. A section where the input signal level has an amplitude greater than or equal to a predetermined threshold is detected, and the time ratio is calculated. If this ratio is high, the strain is large, and if the ratio is small, the strain is small. If the amplitude does not exceed the threshold, no distortion can be achieved. The threshold is set according to the clipping level of the microphone element, circuit, and AD conversion.

<Sound Collection Condition Extraction Unit 11 (S11), Selection Unit 12 (S12)>
Hereinafter, an example of the operation (S11, S12) of the sound collection condition extraction unit 11 and the selection unit 12 will be described. The sound collection condition extraction unit 11 extracts, for example, a feature amount representing the sound collection condition from the input sound signal, and the sound signal input according to the value of the feature amount is grouped (for example, a code representing the sound collection condition). Divide.

  Next, as shown in Table 1, the selection unit 12 transmits a corresponding acoustic signal based on the relationship between a group (a code representing a sound collection condition) and an index (a code representing a destination speech recognition server device). A previous voice recognition server device (for example, the voice recognition server device 21-1) is selected (S12).

  The feature amount x is, for example, the S / N ratio that is the ratio of the size of the audio signal included in the acoustic signal and the size of the background noise signal, the presence or absence of distortion of the acoustic signal, the frequency of distortion, the spectrum shape of the background noise signal, the background The magnitude of the noise signal, etc.

When the feature quantity x is an S / N ratio, for example, threshold values are set as θ ₁ = 0 dB, θ ₂ = 10 dB, θ ₃ = 20 dB, and so on. Are extracted as sound collection conditions, and the selection unit 12 selects index 2.

When the feature quantity x is a distortion of an acoustic signal, for example, the ratio of the time when the absolute value of the amplitude is 30000 or more in 0.5 seconds is a feature quantity x in a signal quantized with a bit depth of 16 bits. The threshold is set as θ ₁ = 0.8, etc., and if x = 0, the sound collection condition extraction unit 11 extracts group 1 as the sound collection condition, the selection unit 12 selects index 1, and if x = 0.9 The sound collection condition extraction unit 11 extracts group 2 as the sound collection condition, and the selection unit 12 selects index 2.

When the feature amount x is the spectrum shape of the background noise signal, for example, the size of the background noise signal is x ₁ , x ₂ ,..., X _m (m is an integer satisfying m ≧ 2) depending on the frequency band and duration. Separately evaluate. The sound collection condition extraction unit 11 extracts a group from the combination of evaluation results, and the selection unit 12 selects the index. As another method of using the spectrum shape of the background noise signal as a feature amount, it is also possible to store a plurality of types of background noise signal models and classify the input background noise signal into one of the models. it can. The multiple types of background noise signals are, for example, white noise, pink noise, burst noise, and the like. In this method, a group corresponding to each model is allocated, and a group is determined according to a model in which background noise signals of input signals are classified.

When the feature amount x is set to the size of the background noise signal, for example, θ ₁ = 40 dBA, θ ₂ = 55 dBA, θ ₃ = 70 dBA, etc. are set, and if the feature amount x = 50 dBA, the sound collection condition extraction unit 11 sets the group 2 Extracting is performed as a sound collection condition, and the selection unit 12 selects index 2. Here, dBA is a unit of dB value of noise level measured under frequency weighting characteristics (A characteristics) considering human hearing.

<Signal processing unit 13 (S13)>
The signal processing unit 13 performs signal processing on the corresponding acoustic signal when the extracted sound collection condition corresponds to a predetermined condition (S13). Specifically, the signal processing unit 13 is a speech recognition target in a speech recognition server device in which the S / N ratio and the size of the background noise signal are determined based on the sound collection conditions extracted by the sound collection condition extraction unit 11. The corresponding acoustic signal is subjected to signal processing so as to conform to the range of the feature amount assumed as. For example, in the sound pickup condition in the vicinity of S / N ratio = 1, that is, in the vicinity of 0 dB, the size of the audio signal and the size of the background noise signal are the same. Easy to invite. Accordingly, when the sound collection condition extraction unit 11 extracts a sound collection condition in the vicinity of S / N ratio = 0 dB, the signal processing unit 13 performs signal processing for suppressing the background noise signal with respect to the sound signal of the sound collection condition. Apply. Alternatively, for example, when the sound collection condition extraction unit 11 extracts a sound collection condition in the vicinity of S / N ratio = 100, that is, in the vicinity of 20 dB, according to the value of the S / N ratio, similar to the sound collection condition in the vicinity of 0 dB described above. Thus, the process of adaptively suppressing the background noise signal may be performed, or the process of suppressing may not be performed at all. Even in other sound collection conditions, the signal processing unit 13 adaptively performs signal processing on the acoustic signal based on the result extracted by the sound collection condition extraction unit 11.

Hereinafter, an example of the operation (S13) of the signal processing unit 13 will be described. In speech recognition, in many cases, input speech is corrected by signal processing as preprocessing. Examples of acoustic characteristics to be dealt with by preprocessing in speech recognition include additive noise and multiplicative noise. Additive noise is a signal that is additively observed with respect to a speech signal, such as noise ubiquitous in the speech input environment. On the other hand, multiplicative noise is noise (distortion) caused by acoustic characteristics such as microphone characteristics and spatial transfer characteristics, and is observed as a convolution operation with respect to the original speech waveform in the time waveform. It becomes a distortion. Examples of speech recognition processing that addresses additive noise include the noise suppression method based on the spectral subtraction method disclosed in paragraph [0005] of Reference Patent Document 1, or the Wiener method disclosed in paragraph [0007] of that document. There is a method of suppressing noise from speech with superimposed noise and applying it to speech recognition, such as a noise suppression method based on a filter method (hereinafter referred to as WF method).
(Reference Patent Document 1: Japanese Patent No. 4464797)

As an example of speech recognition processing that copes with multiplicative noise in addition to additive noise, a noise superimposed speech model is generated by superimposing a noise model on a speech model from which the influence of multiplicative noise has been removed as in Patent Document 1. Then, there is a method of updating the model based on the multiplicative feature amount. Alternatively, there is a method of generating a normalized noise superimposed speech model after normalizing a noise model based on the multiplicative noise feature quantity as in the invention of Reference Patent Document 2.
(Reference Patent Document 2: Japanese Patent No. 5200080)

  Noise suppression is typically considered as signal processing performed by the signal processing unit 13. As signal processing other than noise suppression, for example, AGC (Automatic Gain Control), CMN (Cepstrum Mean Normalization), and an equalizer may be used.

<AGC>
Automatic Gain Control (AGC) detects the input signal level based on the short-time average power or short-time average amplitude of the input audio signal so that the difference between the input signal level and the optimum level (target value) is reduced. This is a process for adjusting the gain of the audio input stage. AGC has an effect of preventing the voice feature amount from becoming unclear due to the voice waveform after A / D conversion being too small or excessive. AGC is disclosed in paragraph [0001] of Reference Patent Document 3, for example.
(Reference Patent Document 3: Japanese Patent No. 3588555)

<CMN>
Cepstrum Mean Normalization (CMN) is a process of obtaining a long-term cepstrum average of an input speech signal in a cepstrum that is a feature amount of speech recognition, and subtracting the long-term cepstrum average from the cepstrum of the input speech of each frame. The CMN is used to reduce the influence of multiplicative distortion represented by the characteristics of the microphone, the position of the microphone, and the shape of the room. CMN is disclosed in paragraph [0010] of Reference Patent Document 1, for example.

  When the CMN is performed by the signal processing unit 13 of the client device 10, the client device 10 transmits to the speech recognition server device an MFCC (Mel Frequency Cepstrum after CMN application) as a signal derived from an acoustic signal for speech recognition. ) Is transmitted, it is possible to omit the cepstrum analysis process again by the speech recognition server device.

<Equalizer>
The equalizer is a process for adjusting the gain of the input audio signal for each frequency band. For example, if it is known in advance that the acoustic characteristics of a microphone for voice input are not flat, sound can be collected after improving the acoustic characteristics via an equalizer. The equalizer is disclosed in, for example, paragraphs [0010] and [0016] of Reference Patent Document 4.
(Reference Patent Document 4: Japanese Patent No. 2865268)

  Next, the transmission unit 14 transmits an acoustic signal or a signal derived from the acoustic signal to the speech recognition server device (speech recognition server device selected in step S12) corresponding to the extracted sound collection condition (S14). . At this time, the transmission unit 14 varies the transmission destination between the case where the signal processing of step S13 is not performed and the case where the signal processing is performed, and the acoustic signal which is not subjected to signal processing or the acoustic signal which is subjected to signal processing. A signal shall be transmitted. Moreover, you may determine the transmission destination in any one of different speech recognition server apparatuses only by whether the signal processing of step S13 was implemented irrespective of the speech recognition server apparatus selected by step S12. Note that the signal derived from the acoustic signal refers to a signal representing a feature amount of the acoustic signal, an acoustic signal subjected to the signal processing in step S13, and the like. In addition, when the transmission unit 14 transmits an acoustic signal or a signal derived from the acoustic signal, the sound recognition condition (group), a threshold value thereof, and information on the presence or absence of signal processing in the signal processing unit 13 are transmitted to the voice recognition server device. You may send it. The voice recognition server device can record the sound pickup condition or signal processing condition for selecting the voice recognition server device from the sound pickup condition (group), its threshold value, and the presence or absence of signal processing. become.

  The voice recognition server devices 21-1, ..., 21-n, ..., 21-N receive an acoustic signal or a signal derived from the acoustic signal from the client device 10 (S21A). The speech recognition server device (for example, the speech recognition server device 21-1) that has received the acoustic signal or the signal derived from the acoustic signal executes speech recognition processing (S21B).

<Voice recognition processing (S21B)>
The voice recognition process in step S21B is executed as follows, for example. The speech recognition server device converts an utterance of one sentence or one word into a character string. The speech recognition server device uses speech power and its variation, MFCC (Mel-Frequency Cepstrum Coefficient) and its dynamic variation as speech feature amounts. The speech recognition server device searches for a word string using a statistical acoustic model or a language model.

  The voice recognition server apparatus that has executed the voice recognition process in step S21B transmits the voice recognition result to the client apparatus 10 (S21C). The receiving unit 15 of the client device 10 receives the voice recognition result (S15A). The presentation unit 16 of the client device 10 presents the received voice recognition result (S16).

  Hereinafter, the information updating operation of the voice recognition system 1 of the present embodiment will be described with reference to FIGS. FIG. 5 is a sequence diagram showing an information update operation of the speech recognition system 1 of the present embodiment. FIG. 6 is a flowchart showing the operation of the rephrase determining unit 17 of the voice recognition system 1 of the present embodiment. The rephrasing determination unit 17 monitors and acquires a repeated signal group, which is a signal group in which a plurality of repetitions of utterances indicating the same content by the user are observed (S17). In order to recognize a plurality of repetitions of utterances indicating the same content by the user, each constituent requirement (see FIG. 2) of the rephrase determining unit 17 executes, for example, the following processing (see FIG. 6). Here, it is assumed that a total of M acoustic signals (M is an integer satisfying M ≧ 2) are input to the client device 10, and m is an integer satisfying 2 ≦ m ≦ M. A case will be described in which it is determined whether or not the m-th acoustic signal is rephrased according to each component requirement.

  The reaction time measurement unit 17A includes a time when the client device 10 presents a speech recognition result for the acoustic signal input to the client device 10 (hereinafter referred to as the (m-1) th presentation time) and the client device 10. The reaction time, which is the difference from the input time of the mth input acoustic signal (hereinafter referred to as the mth input time), is measured (S17A).

  The reliability acquisition unit 17B acquires the reliability of the speech recognition result (hereinafter referred to as the m-1th speech recognition result) for the acoustic signal input to the client device 10 as the m-1st (S17B). The reliability is information transmitted from the voice recognition server device to the client device 10 together with character string information indicating the voice recognition result.

  The similarity calculation unit 17 </ b> C includes the similarity of each of the m−1 and mth input acoustic signals to the client device 10, and each sound corresponding to each of the m−1 and mth input acoustic signals to the client device 10. At least one of the similarities of the recognition results is calculated (S17C).

  Whether the determination unit 17D rephrases the m-th acoustic signal input to the client device 10 based on at least one of the reaction time, reliability, and similarity acquired in steps S17A to S17C. It is determined whether or not, and a repeated signal group is acquired based on the determination result (S17D). The determination unit 17D determines that the user's reaction time is smaller than a predetermined threshold, the reliability of the (m-1) th speech recognition result is lower than the predetermined threshold, and the similarity is higher than the predetermined threshold. , The (m-1) th speech recognition result is misrecognition, and the utterance included in the mth acoustic signal is determined to be rephrased by the user (repetition of the same utterance). That is, the determination unit 17D acquires the m−1th input acoustic signal and the mth input acoustic signal as a repeated signal group.

  Note that steps S17A to S17C are not necessarily executed in the order described above, and the order of these processes may be changed. For example, if steps S17A to S17C are to be executed starting with a light processing load, step S17B is executed first, step S17A is executed next, and step S17C is finally executed. Further, after each step of steps S17A to S17C, the determination of step S17D may be performed one by one, and if it is determined that it is not rephrased, the subsequent steps may be aborted. As described above, among the steps S17A to S17C, the process having the lightest processing load is executed first (S17B → S17A → S17C), and when it is determined in the first step (S17B) that it is not rephrased, the remaining Two steps (S17A, S17C) are omitted, and when it is determined that the second step (S17A) is not rephrased, the remaining one step (S17C) is omitted, thereby reducing the load on the client device 10. Can be reduced.

The similarity in step S17C can be, for example, one of the following.
A value obtained by inverting the reciprocal value or sign of the Euclidean distance of the feature value of each acoustic signal (cepstrum, power, or a change amount thereof can be used as the feature value).
A value obtained by inverting the reciprocal value or sign of the edit distance of the character string of each voice recognition result obtained from the voice recognition server device. Here, the character string is not limited to a character string on the notation, and may be a character string obtained by reading a character string and converting it into a kana or phoneme notation.

  Note that the speech recognition server apparatus selected in step S12 executes speech recognition for the repeated signal group (S21A to S21C).

  The reason for using the above reaction time for the rephrase determination is that in the case of rephrase, the time from the presentation of the previous recognition result to the next utterance tends to be shorter than in the case where it is not. Because it is recognized. The reason why the reliability is used for the re-statement determination is that when the recognition result is incorrect, the reliability of the speech recognition result tends to be low. The reason why the similarity is used for the re-statement determination is that, in the case of re-statement, a tendency that the degree of similarity between utterances increases is recognized. Since the reaction time of the user is the time until the user reads and understands the recognition result, the threshold for determining the reaction time may be a value proportional to the number of characters presented as the recognition result. If the recognition result includes kanji, the threshold may be longer depending on the number of kanji characters.

  The rephrase determination unit 17 (determination unit 17D) extracts the speech recognition result of the last signal from the repeated signal group as a correct candidate (S17). The rephrase determination unit 17 (determination unit 17D) sets all the repetitive signal groups as relearning signal groups, and stores them in the reword information storage unit 171 in association with correct answer candidates. The reason for using all repetitive signal groups as the relearning signal group is that the speech recognition server device that can generate a speech recognition result that is equal to the correct answer candidate or has a high similarity with the correct answer candidate in all of the repetitive signal groups, This is because it is suitable as a transmission destination.

  Hereinafter, an example of the restatement determination operation of the restatement determination unit 17 will be described with reference to FIG. FIG. 7 is a diagram illustrating the rephrasing determination operation of the voice recognition system 1 according to the present embodiment. Although the speech recognition system 1 of the present embodiment enables speech recognition of sentences not limited to words, an example of word speech recognition will be described for easy understanding of the main points. As shown in FIG. 7, it is assumed that the user 9 of the client apparatus 10 utters “Kiryu” (Kiryu) toward the client apparatus 10 (hereinafter, this utterance is referred to as “utterance 1”). The client device 10 transmits an acoustic signal including the utterance 1 to the voice recognition server device (here, the voice recognition server device 21-n) selected in step S12. The speech recognition server device 21-n recognizes an acoustic signal including the utterance 1 and returns a speech recognition result “Chiryu” to the client device 10 (hereinafter, the speech recognition result is referred to as a recognition result 1). . The client device 10 presents the recognition result 1 to the user 9.

  The user 9 notices that the presented recognition result 1 is a misrecognition, and utters “Kiryu” (Kiryu) toward the client device 10 in the same manner as before (hereinafter, this utterance is referred to as utterance 2). Shall. The client device 10 transmits an acoustic signal including the utterance 2 to the voice recognition server device 21-n. The voice recognition server device 21-n recognizes an acoustic signal including the utterance 2 and returns a voice recognition result “Kiryu” to the client device 10 (hereinafter, the voice recognition result is referred to as a recognition result 2). . The client device 10 presents the recognition result 2 to the user 9.

  The user 9 looks at the presented recognition result 2 and confirms that the voice recognition has been executed correctly. This time, the user 9 is directed to the client device 10 and speaks (hereinafter referred to as utterance 3). ). The client device 10 transmits an acoustic signal including the utterance 3 to the voice recognition server device 21-n. The voice recognition server device 21-n recognizes an acoustic signal including the utterance 3 and returns a voice recognition result “Kudo” to the client device 10 (hereinafter, the voice recognition result is referred to as a recognition result 3). The client device 10 presents the recognition result 3 to the user 9.

  In the above example, the reaction time measurement unit 17A measures the reaction time (hereinafter referred to as reaction time 1) that is the difference between the presentation time of the recognition result 1 and the input time of the acoustic signal including the utterance 2 (S17A). The reliability acquisition unit 17B acquires the reliability of the recognition result 1 (S17B). The similarity calculation unit 17C calculates at least one of the similarity between the acoustic signal including the utterance 1 and the acoustic signal including the utterance 2 or the similarity between the recognition result 1 and the recognition result 2 (S17C). In this case, the determination unit 17D determines that the acoustic signal including the utterance 1 and the acoustic signal including the utterance 2 have the reaction time 1 smaller than the predetermined threshold, and the reliability of the recognition result 1 is lower than the predetermined threshold. Since the similarity or the similarity between the recognition result 1 and the recognition result 2 is higher than a predetermined threshold, the recognition result 1 is a false recognition, and the utterance 2 is a rephrase by the user (repetition of the same utterance). (S17D).

  Similarly, the determination unit 17D does not make the reaction time 2 smaller than the predetermined threshold (the reaction time 2 is sufficiently long), and the reliability of the recognition result 2 does not become lower than the predetermined threshold (the recognition result 2). The degree of reliability is sufficiently high), the similarity between the acoustic signal including the utterance 2 and the acoustic signal including the utterance 3, or the similarity between the recognition result 2 and the recognition result 3 does not become higher than a predetermined threshold (two acoustic signals, It is determined that the recognition result 2 is a correct candidate and the utterance 3 is not re-stated by the user (repetition of the same utterance) (S17D). In this case, the acoustic signal including the utterance 1 and the acoustic signal including the utterance 2 correspond to the relearning signal group.

  Next, the transmission unit 14 transmits a set of the correct answer candidate and the relearning signal group to the management unit 30 (S14B).

  The correct answer candidate receiving unit 30A of the management unit 30 receives a set of a correct answer candidate and a relearning signal group from the client device 10 (S30A). A set of the correct answer candidate and the relearning signal group is stored in the correct answer candidate storage unit 30F. The relearning signal group transmission unit 30B of the management unit 30 transmits the relearning signal group to the speech recognition server device group 20 (all speech recognition server devices) (S30B).

  The voice recognition server device group 20 receives the relearning signal group from the management unit 30 (S21D). The speech recognition server device group 20 recognizes the received relearning signal group (S21E). The voice recognition server device group 20 transmits the voice recognition result to the management unit 30 (S21F).

  The speech recognition result receiving unit 30C of the management unit 30 receives speech recognition results for the relearning signal group from all speech recognition server devices (S30C). The transmission destination information update unit 30D of the management unit 30 updates the transmission destination information based on the similarity between each voice recognition result received from all voice recognition server devices and the correct candidate (S30D). The transmission destination information is information relating to the relationship between the voice recognition server device that is the transmission destination of the acoustic signal and the sound collection conditions. Typically, the transmission destination information update unit 30D performs similarity (first similarity, second similarity) of the speech recognition result for the correct answer candidate and the relearning signal group (L signal group, L ≧ 2). ,..., The Lth similarity) is increased, or the acoustic signal from the client device 10 is transmitted to the speech recognition server device in which the reliability of the speech recognition result for the relearning signal group is increased. As described above, the destination information is updated (S30D). In other words, in step S30D, a speech recognition server device that can generate a speech recognition result equal to the correct answer (high similarity) for any relearning signal in the relearning signal group is a suitable transmission destination. Selected. For example, when the above-mentioned utterance “Kiryu” (Kiryu) is re-stated L times, a total of L relearning signals including the utterance “Kiryu” (Kiryu) exist. At this time, the preferred speech recognition server device of the transmission destination is equal to “Kiryu” (Kiryu) which is a correct answer candidate for any of the relearning signal groups including L utterances “Kiryu” (Kiryu). Or a speech recognition server device that can generate a speech recognition result with a high degree of similarity. In step S30D, a speech recognition server device that can generate a highly reliable speech recognition result for any relearning signal in the relearning signal group may be selected as a suitable transmission destination. In the above example, a voice recognition server device capable of generating a highly reliable voice recognition result for any of the relearning signal groups including L utterances “Kiryu” (Kiryu) is a suitable destination. May be selected. The transmission destination information update unit 30D determines the number of times that the similarity (first similarity, second similarity,..., Lth similarity) of the speech recognition result for the correct candidate and the relearning signal group is greater than a predetermined threshold. The acoustic signal from the client device 10 is transmitted to the most speech recognition server device or the speech recognition server device having the largest number of times the reliability of the speech recognition result for the relearning signal group is greater than a predetermined threshold. As described above, the destination information may be updated (S30D). In other words, in step S30D, a speech recognition server device that can generate the largest number of speech recognition results that are equal to the correct candidate (high similarity) for the relearning signal group is selected as a suitable transmission destination. For example, when the above-mentioned utterance “Kiryu” (Kiryu) is re-stated L times, a total of L relearning signals including the utterance “Kiryu” (Kiryu) exist. At this time, the preferred speech recognition server device of the transmission destination is equal to or higher than “Kiryu” which is a correct answer candidate for the relearning signal group including L utterances “Kiriu” (Kiryu). This is a voice recognition server device that can generate the most similar voice recognition results. In step S30D, a speech recognition server device that can generate the most highly reliable speech recognition results for the relearning signal group may be selected as a suitable transmission destination. In the above example, a speech recognition server device that can generate the most highly reliable speech recognition results for a re-learning signal group including L utterances “Kiryu” (Kiryu) is a suitable transmission destination. It may be selected.

  Hereinafter, an example of the transmission destination information update operation of the management unit 30 will be described with reference to FIG. FIG. 8 is a diagram illustrating the transmission destination information update operation of the speech recognition system 1 of the present embodiment. As illustrated in FIG. 8, the client device 10 transmits the acoustic signal including the utterance 1 that is the re-learning signal group, the acoustic signal including the utterance 2, and the recognition result 2 that is the correct candidate to the management unit 30. . The management unit 30 transmits the acoustic signal including the utterance 1 and the acoustic signal including the utterance 2 that are the relearning signal group to the speech recognition server device 21-b and the speech recognition server device 21-c. It is assumed that the speech recognition server device 21-b returns the recognition results 1b and 2b to the acoustic signal including the utterance 1 and the acoustic signal including the utterance 2. It is assumed that the recognition results 1b and 2b include at least one “chiryu” that is a misrecognition. On the other hand, the speech recognition server device 21-c returns the recognition results 1c and 2c to the acoustic signal including the utterance 1 and the acoustic signal including the utterance 2, and both the recognition results 1c and 2c are correct candidates. "Kiryu", which is the same result as

  In the example of FIG. 8, since the recognition result 2 (correct answer candidate) and the recognition results 1c and 2c are equal and the reliability of the recognition results 1c and 2c is equal to or higher than a predetermined threshold, the management unit 30 recognizes the voice recognition server device 21. -C is the optimal transmission destination, and the recognition setting (referred to as recognition setting C) and the acoustic model held by the device are determined to be optimal for the corresponding acoustic signal, and the speech recognition server device The above-mentioned transmission destination information is updated so that 21-c becomes the transmission destination of the acoustic signal.

  Next, the transmission destination information transmission unit 30E of the management unit 30 transmits the updated transmission destination information to the client device 10 (S30E).

  The receiving unit 15 of the client device 10 receives the transmission destination information from the management unit 30 (S15B). The transmission destination changing unit 18 of the client device 10 changes the relationship between the voice recognition server device that is the transmission destination of the acoustic signal and the sound collection conditions based on the received transmission destination information (S18). The transmission destination changing unit 18 of the client device 10 can perform the above-described change by overwriting the transmission destination information stored in the transmission destination storage unit 121 with the newly received transmission destination information, for example.

  As described above, according to the speech recognition system 1 of the present embodiment, all speech recognition server devices recognize the relearning signal group as speech, and are equal to the correct candidate for any relearning signal group (high similarity). ) The management unit 30 sets the voice recognition server apparatus that has transmitted the voice recognition result, or the voice recognition server apparatus that has transmitted the voice recognition result with high reliability for any relearning signal group, as a new transmission destination. The destination information is updated so that the number of misrecognitions in the entire system decreases by changing the voice recognition server device as the destination based on the destination information updated by the client device 10. Is optimized (relearning), so that the utilization efficiency of the system can be improved.

  Hereinafter, the voice recognition system according to the second embodiment that achieves the same effect as the first embodiment by updating the threshold value instead of updating the transmission destination information will be described. First, the configuration of the speech recognition system according to the present embodiment will be described with reference to FIGS. FIG. 9 is a block diagram showing the configuration of the voice recognition system 2 of the present embodiment. FIG. 10 is a block diagram illustrating the configuration of the management unit 50 of the voice recognition system 2 of the present embodiment. As shown in FIG. 9, the speech recognition system 2 according to the present exemplary embodiment includes a client device 40, a plurality of speech recognition server devices 21-1 to 21 -n, 21 to N, and a management unit 50. . There may be a plurality of client devices 40. It is assumed that the client device 40 and the voice recognition server device group 20 are connected to be communicable wirelessly or by wire via a network. The management unit 50 may be configured as a single hardware (device). When the management unit 50 is configured as a single hardware (device), this may be referred to as the management device 50. When the management unit 50 is configured as a single piece of hardware (device), the client device 40, the voice recognition server device group 20, and the management unit 50 (management device 50) are connected to be communicable wirelessly or wired via a network. It shall be. Further, the management unit 50 may be a configuration requirement in the client device 40 or a configuration requirement in any of the voice recognition server devices in the voice recognition server device group 20.

  As illustrated in FIG. 9, the client device 40 according to the present exemplary embodiment includes a threshold value changing unit 48 instead of the transmission destination changing unit 18 included in the client device 10 according to the first exemplary embodiment. Since the configuration requirements of the client device 40 other than the threshold value changing unit 48 are the same as the configuration requirements of the client device 10 of the first embodiment, a description thereof will be omitted.

  As shown in FIG. 10, the management unit 50 of the present embodiment includes a transmission destination information update unit 30D included in the management unit 30 of the first embodiment, a threshold value update unit 50D instead of the transmission destination information transmission unit 30E, A threshold transmission unit 50E is provided. Further, the management unit 50 according to the present embodiment includes a threshold value storage unit 50G that is not included in the management unit 30 according to the first embodiment, and a signal processing unit 50H. Since the configuration requirements other than the threshold update unit 50D, the threshold transmission unit 50E, the threshold storage unit 50G, and the signal processing unit 50H are the same as the configuration requirements of the management unit 30 of the first embodiment, a description thereof is omitted. To do.

  Note that the voice recognition operation of the voice recognition system 2 of the present embodiment is completely the same as the voice recognition operation (S11 to S14A, S21A to S21C, S15A, and S16) of the first embodiment, and thus description thereof is omitted.

  Hereinafter, the information update operation of the voice recognition system 2 of the present embodiment will be described with reference to FIG. FIG. 11 is a sequence diagram showing an information update operation of the speech recognition system 2 of the present embodiment.

  Steps S17, S14B, and S30A are executed in the same manner as in the first embodiment. Next, the signal processing unit 50H is the signal processing executed in step S13 described above, and includes all of the signal processing patterns predetermined according to the sound collection conditions (including patterns in which the signal processing itself is not performed). ) Is applied to the acoustic signal before the signal processing in step S13 is executed, and a signal-processed acoustic signal subjected to different signal processing for each sound collection condition is acquired. In the present embodiment, this signal processed acoustic signal is set as a relearning signal group (S50H).

  Note that if an increase in load due to the signal processing in step S50H becomes a problem, step S50H may be omitted. In this case, the signal processing applied to the relearning signal group may be different from the signal processing applied to the acoustic signal in step S13 after the threshold value change described later, and the signal processing is performed in step S13. The acoustic signal is directly used as a relearning signal group. Step S30B is the same as that of the first embodiment. However, when step S50H is executed, each acoustic signal (each relearning signal) subjected to each signal processing in S50H according to each sound collection condition is It is assumed that it is distributed to each voice recognition server device in charge of the corresponding sound collection condition. Step S30B in this case can be said to be processing in which steps S13 to S14 in all sound collection condition patterns are reproduced by the management unit 50.

  Hereinafter, Steps S21D to S21F and Step S30C are executed in the same manner as in the first embodiment.

  Next, the threshold update unit 50D updates the threshold based on the similarity between each speech recognition result received from all speech recognition server devices and the correct answer candidate or the reliability of each speech recognition result. (S50D). As described above, the threshold value is a value set in advance for extracting the sound collection condition. Typically, the threshold update unit 50D increases the similarity (first similarity, second similarity,..., Lth similarity) of the speech recognition results for the correct candidate and the relearning signal group. Alternatively, the threshold value is updated so that the acoustic signal from the client device 10 is transmitted to the speech recognition server device in which the reliability of the speech recognition result for the relearning signal group is high (S50D). ). Next, the threshold transmission unit 50E of the management unit 50 transmits the updated threshold to the client device 40 (S50E).

  The receiving unit 15 of the client device 40 receives the threshold value from the management unit 50 (S15B). The threshold value changing unit 48 of the client device 40 changes a preset threshold value based on the received threshold value (S48). The threshold value changing unit 48 of the client device 40 can make the above-described change by overwriting the threshold value already stored in the threshold value storage unit 111 with the newly received threshold value, for example.

  As described above, according to the speech recognition system 2 of the present embodiment, all speech recognition server devices recognize a relearning signal group as speech, and are equal to a correct candidate for any relearning signal group (high similarity). ) The management unit 50 so that the voice recognition server apparatus that has transmitted the voice recognition result or the voice recognition server apparatus that has transmitted the voice recognition result having high reliability for any relearning signal group becomes a new transmission destination. Updates the threshold value, and the client device 40 changes the preset threshold value to a new threshold value based on the updated threshold value, thereby reducing the number of false recognitions in the entire system. Since the threshold value is optimized (re-learning) in the direction, the utilization efficiency of the system can be improved.

  Hereinafter, instead of changing the voice recognition server device as the transmission destination, the voice recognition system according to the third embodiment configured to update (replace, relearn) settings related to the acoustic model and the voice recognition stored in the voice recognition server device. Will be described. First, the configuration of the voice recognition system according to the present embodiment will be described with reference to FIGS. FIG. 12 is a block diagram showing the configuration of the voice recognition system 3 of the present embodiment. FIG. 13 is a block diagram illustrating a configuration of the management unit 90 of the voice recognition system 3 according to the present embodiment. As shown in FIG. 12, the voice recognition system 3 of the present embodiment includes a client device 70, a plurality of voice recognition server devices 81-1,..., 81-n,. . .., 81 -N are collectively referred to as a voice recognition server device group 80. There may be a plurality of client devices 70. It is assumed that the client device 70 and the voice recognition server device group 80 are connected via a network so that they can communicate wirelessly or by wire. The management unit 90 may be configured as a single hardware (device). When the management unit 90 is configured as a single hardware (device), this may be referred to as a management device 90. When the management unit 90 is configured as a single piece of hardware (device), the client device 70, the voice recognition server device group 80, and the management unit 90 (management device 90) are connected to be communicable wirelessly or wired via a network. It shall be. Further, the management unit 90 may be a configuration requirement in the client device 70 or a configuration requirement in any voice recognition server device in the voice recognition server device group 80.

  As illustrated in FIG. 12, the client device 70 according to the present exemplary embodiment does not include the transmission destination changing unit 18 included in the client device 10 according to the first exemplary embodiment. Since the other configuration requirements of the client device 70 are the same as the configuration requirements of the client device 10 of the first embodiment, a description thereof will be omitted.

  As illustrated in FIG. 13, the management unit 90 according to the present embodiment includes a transmission destination information update unit 30D included in the management unit 30 according to the first embodiment, a setting information update unit 90D instead of the transmission destination information transmission unit 30E, An information transmission unit 90E is provided. Since the configuration requirements other than the setting information update unit 90D and the setting information transmission unit 90E are the same as the configuration requirements of the management unit 30 of the first embodiment, a description thereof will be omitted.

  Note that the voice recognition operation of the voice recognition system 3 of the present embodiment is completely the same as the voice recognition operation (S11 to S14A, S21A to S21C, S15A, and S16) of the first embodiment, and thus the description thereof is omitted.

  Hereinafter, the information update operation of the speech recognition system 3 of the present embodiment will be described with reference to FIG. FIG. 14 is a sequence diagram showing the information update operation of the speech recognition system 3 of the present embodiment. Steps S17 to S14B, S30A to S30B, S21D to S21F, and S30C are executed in the same manner as in the first embodiment.

  Next, the setting information update unit 90D of the management unit 90, in step S12, based on the similarity between each speech recognition result received from all speech recognition server devices and correct answer candidates or the reliability of each speech recognition result. The setting information of the selected voice recognition server device is updated (S90D). The setting information is information related to voice recognition settings, and includes information specifying an acoustic model and information specifying settings related to voice recognition. The setting information may include the acoustic model itself. Typically, the setting information update unit 90D increases the similarity (first similarity, second similarity,..., Lth similarity) of the speech recognition results for the correct candidate and the relearning signal group, or Update the setting information of the voice recognition server device selected in step S12 so that the same voice recognition setting and the same acoustic model as the voice recognition server device in which the reliability of the voice recognition result for the re-learning signal group becomes high. (S90D). The setting information transmission unit 90E of the management unit 90 transmits the updated setting information to the voice recognition server device selected in step S12 (S90E).

  The voice recognition server device selected in step S12 receives the setting information (S80G), and changes its own voice recognition setting (acoustic model, settings related to voice recognition) based on the received setting information (S80H). ). The voice recognition server device group 80 may change the setting in a time zone (for example, at night) when the operation amount of the voice recognition server device group 80 is small. In addition, the voice recognition server device group 80 may change the setting in other time zones planned in advance.

  Thus, according to the voice recognition system 3 of the present embodiment, the setting information update unit 90D of the management unit 90 updates the setting information, and the voice recognition server device selected in step S12 is updated based on the updated setting information. Since the settings relating to the acoustic model and speech recognition of the own device are changed, the erroneous recognition of the speech recognition server device selected in step S12 can be reduced, and the utilization efficiency of the system can be improved.

<Supplementary note>
The apparatus of the present invention includes, for example, a single hardware entity as an input unit to which a keyboard or the like can be connected, an output unit to which a liquid crystal display or the like can be connected, and a communication device (for example, a communication cable) capable of communicating outside the hardware entity. Can be connected to a communication unit, a CPU (Central Processing Unit, may include a cache memory or a register), a RAM or ROM that is a memory, an external storage device that is a hard disk, and an input unit, an output unit, or a communication unit thereof , A CPU, a RAM, a ROM, and a bus connected so that data can be exchanged between the external storage devices. If necessary, the hardware entity may be provided with a device (drive) that can read and write a recording medium such as a CD-ROM. A physical entity having such hardware resources includes a general-purpose computer.

  The external storage device of the hardware entity stores a program necessary for realizing the above functions and data necessary for processing the program (not limited to the external storage device, for example, reading a program) It may be stored in a ROM that is a dedicated storage device). Data obtained by the processing of these programs is appropriately stored in a RAM or an external storage device.

  In the hardware entity, each program stored in an external storage device (or ROM or the like) and data necessary for processing each program are read into a memory as necessary, and are interpreted and executed by a CPU as appropriate. . As a result, the CPU realizes a predetermined function (respective component requirements expressed as the above-described unit, unit, etc.).

  The present invention is not limited to the above-described embodiment, and can be appropriately changed without departing from the spirit of the present invention. In addition, the processing described in the above embodiment may be executed not only in time series according to the order of description but also in parallel or individually as required by the processing capability of the apparatus that executes the processing. .

  As described above, when the processing functions in the hardware entity (the apparatus of the present invention) described in the above embodiments are realized by a computer, the processing contents of the functions that the hardware entity should have are described by a program. Then, by executing this program on a computer, the processing functions in the hardware entity are realized on the computer.

  The program describing the processing contents can be recorded on a computer-readable recording medium. As the computer-readable recording medium, for example, any recording medium such as a magnetic recording device, an optical disk, a magneto-optical recording medium, and a semiconductor memory may be used. Specifically, for example, as a magnetic recording device, a hard disk device, a flexible disk, a magnetic tape or the like, and as an optical disk, a DVD (Digital Versatile Disc), a DVD-RAM (Random Access Memory), a CD-ROM (Compact Disc Read Only). Memory), CD-R (Recordable) / RW (ReWritable), etc., magneto-optical recording medium, MO (Magneto-Optical disc), etc., semiconductor memory, EEP-ROM (Electronically Erasable and Programmable-Read Only Memory), etc. Can be used.

  The program is distributed by selling, transferring, or lending a portable recording medium such as a DVD or CD-ROM in which the program is recorded. Furthermore, the program may be distributed by storing the program in a storage device of the server computer and transferring the program from the server computer to another computer via a network.

  A computer that executes such a program first stores, for example, a program recorded on a portable recording medium or a program transferred from a server computer in its own storage device. When executing the process, the computer reads a program stored in its own recording medium and executes a process according to the read program. As another execution form of the program, the computer may directly read the program from a portable recording medium and execute processing according to the program, and the program is transferred from the server computer to the computer. Each time, the processing according to the received program may be executed sequentially. Also, the program is not transferred from the server computer to the computer, and the above-described processing is executed by a so-called ASP (Application Service Provider) type service that realizes the processing function only by the execution instruction and result acquisition. It is good. Note that the program in this embodiment includes information that is used for processing by an electronic computer and that conforms to the program (data that is not a direct command to the computer but has a property that defines the processing of the computer).

  In this embodiment, a hardware entity is configured by executing a predetermined program on a computer. However, at least a part of these processing contents may be realized by hardware.

Claims (9)

A voice recognition system including a client device, a plurality of voice recognition server devices, and a management unit,
The client device is
A receiving unit that receives a speech recognition result for an acoustic signal input to the client device from a speech recognition server device selected based on the sound collection condition;
A rephrase determining unit that acquires a repeated signal group that is a signal group obtained by observing a plurality of repetitions of utterances indicating the same content by a user, and extracts a speech recognition result of the last signal from the repeated signal group as a correct candidate When,
All the repetitive signal groups are re-learning signal groups, the correct candidate, and a transmission unit that transmits a set of the re-learning signal groups to the management unit,
Based on the destination information that is the relationship between the voice recognition server device that is the transmission destination of the acoustic signal and the sound collection condition, the voice recognition server device that is the destination of the acoustic signal and the sound collection condition Including a destination change section that changes the relationship,
The management unit
A speech recognition result receiving unit for receiving speech recognition results for the relearning signal group from all speech recognition server devices;
A destination information update unit that updates the destination information based on the similarity between each voice recognition result received from all the voice recognition server devices and the correct candidate;
A speech recognition system including a transmission destination information transmission unit that transmits the updated transmission destination information to the client device. A voice recognition system including a client device, a plurality of voice recognition server devices, and a management unit,
The client device is
A receiving unit that receives a speech recognition result for an acoustic signal input to the client device from a speech recognition server device selected based on the sound collection condition;
A rephrase determining unit that acquires a repeated signal group that is a signal group obtained by observing a plurality of repetitions of utterances indicating the same content by a user, and extracts a speech recognition result of the last signal from the repeated signal group as a correct candidate When,
The repetitive signal group is a re-learning signal group, and includes a transmission unit that transmits a set of the correct answer candidate and the re-learning signal group to the management unit,
The management unit
A speech recognition result receiving unit for receiving speech recognition results for the relearning signal group from all speech recognition server devices;
Setting information for updating setting information, which is information related to the voice recognition setting of the selected voice recognition server device, based on the similarity between each voice recognition result received from all the voice recognition server devices and the correct answer candidate Update section,
A setting information transmitting unit that transmits the updated setting information to the selected voice recognition server device;
Each of the voice recognition server devices
When receiving the setting information, the voice recognition system changes the voice recognition setting of the own device based on the received setting information. A voice recognition system including a client device, a plurality of voice recognition server devices, and a management unit,
The client device is
A receiving unit that receives a speech recognition result for an acoustic signal input to the client device from a speech recognition server device selected based on the sound collection condition;
A rephrase determining unit that acquires a repeated signal group that is a signal group obtained by observing a plurality of repetitions of utterances indicating the same content by a user, and extracts a speech recognition result of the last signal from the repeated signal group as a correct candidate When,
All the repetitive signal groups are re-learning signal groups, the correct candidate, and a transmission unit that transmits a set of the re-learning signal groups to the management unit,
A threshold value changing unit that changes a threshold value that is a preset value for extracting the sound pickup condition;
The management unit
A speech recognition result receiving unit for receiving speech recognition results for the relearning signal group from all speech recognition server devices;
A threshold update unit that updates the threshold based on the similarity between each voice recognition result received from all the voice recognition server devices and the correct answer candidate;
A speech recognition system including a threshold value transmission unit that transmits the updated threshold value to the client device. The speech recognition system according to any one of claims 1 to 3,
The rephrase determining unit
m is an integer greater than or equal to 2, and the time when the speech recognition result for the m-1st acoustic signal input to the client device is presented by the client device and the input of the mth acoustic signal input to the client device The reaction time that is the difference from the time, the reliability of the speech recognition result for the m−1th acoustic signal input to the client device, the m−1th and mth acoustic signal input to the client device The mth input to the client device is based on at least one of the similarity and the similarity of each speech recognition result for each of the m−1th and mth input acoustic signals to the client device. A speech recognition system that determines whether or not the acoustic signal is a rephrase and acquires the repetitive signal group based on the determination result. A voice recognition method executed by a client device, a plurality of voice recognition server devices, and a management unit,
The client device is
Receiving a speech recognition result for an acoustic signal input to the client device from a speech recognition server device selected based on the sound collection condition;
Obtaining a repetitive signal group that is a signal group in which multiple repetitions of utterances indicating the same content by the user are observed, and extracting the speech recognition result of the last signal of the repetitive signal group as a correct candidate;
All the repetitive signal groups are re-learning signal groups, and the step of transmitting the correct candidate and a set of the re-learning signal groups to the management unit,
The management unit
Receiving speech recognition results for the relearning signal group from all speech recognition server devices;
It is information related to the relationship between the voice recognition server device that is the transmission destination of the acoustic signal and the sound collection condition based on the similarity between each voice recognition result received from all the voice recognition server devices and the correct answer candidate. Updating the destination information;
Performing the step of transmitting the updated destination information to the client device;
The client device is
A speech recognition method for executing a step of changing a relationship between a speech recognition server device serving as a transmission destination of the acoustic signal and the sound collection condition based on the transmission destination information. A voice recognition method executed by a client device, a plurality of voice recognition server devices, and a management unit,
The client device is
Receiving a speech recognition result for an acoustic signal input to the client device from a speech recognition server device selected based on the sound collection condition;
Obtaining a repetitive signal group that is a signal group in which multiple repetitions of utterances indicating the same content by the user are observed, and extracting the speech recognition result of the last signal of the repetitive signal group as a correct candidate;
All the repetitive signal groups are re-learning signal groups, and the step of transmitting the correct candidate and a set of the re-learning signal groups to the management unit,
The management unit
Receiving speech recognition results for the relearning signal group from all speech recognition server devices;
Updating setting information, which is information related to voice recognition settings of the selected voice recognition server device, based on the similarity between each voice recognition result received from all the voice recognition server devices and the correct answer candidate; ,
Executing the step of transmitting the updated setting information to the selected voice recognition server device;
Each of the voice recognition server devices
A voice recognition method for executing a step of changing a voice recognition setting of the own apparatus based on the received setting information when the setting information is received. A voice recognition method executed by a client device, a plurality of voice recognition server devices, and a management unit,
The client device is
Receiving a speech recognition result for an acoustic signal input to the client device from a speech recognition server device selected based on the sound collection condition;
Obtaining a repetitive signal group that is a signal group in which multiple repetitions of utterances indicating the same content by the user are observed, and extracting the speech recognition result of the last signal of the repetitive signal group as a correct candidate;
All the repetitive signal groups are re-learning signal groups, and the step of transmitting the correct candidate and a set of the re-learning signal groups to the management unit,
The management unit
Receiving speech recognition results for the relearning signal group from all speech recognition server devices;
A step of updating a threshold value, which is a value set in advance for extracting the sound pickup condition, based on the similarity between each voice recognition result received from all the voice recognition server devices and the correct answer candidate When,
Performing the step of transmitting the updated threshold to the client device;
The client device is
A speech recognition method for executing the step of changing the threshold value.   A program for causing a computer to function as a voice recognition server device included in the voice recognition system according to any one of claims 1 to 4.   The program for functioning a computer as a management part contained in the speech recognition system in any one of Claim 1 to 4.

Images