Voice recognition device, program, voice recognition method, and voice recognition system

Abstract

Provided is a voice recognition device 1 comprising a voice input unit (10), a voice recognition unit (20), a determination unit (30), a processing execution unit (40), and an adjustment unit (50). The voice input unit (10) receives input of voice of an occupant in a vehicle. The voice recognition unit (20) recognizes the voice input to the voice input unit (10) as first voice information. The determination unit (30) determines whether the similarity between the first voice information and second voice information stored in advance satisfies a predetermined criterion. The processing execution unit (40) executes a predetermined process when the similarity satisfies the predetermined criterion. The adjustment unit (50) adjusts a similarity determination rule and / or the predetermined criterion according to the input from the occupant.

Description

Speech recognition device, program, speech recognition method, and speech recognition system

[0001] The present invention relates to a speech recognition device, a program, a speech recognition method, and a speech recognition system.

[0002] For example, Japanese Patent Laid-Open Publication No. 2000-106592 (Patent Document 1) discloses a telephone device with a voice response function that can effectively prevent malfunction of the voice response function due to ambient noise without impairing the convenience of the voice response function. Also, Japanese Patent Laid-Open Publication No. 2016-85420 (Patent Document 2) discloses a voice adjustment device that appropriately adjusts the input gain and output gain of a voice signal according to the usage environment, thereby improving the voice recognition rate.

[0003] JP 2000-106592 A JP 2016-85420 A

[0004] In vehicles, voice recognition may occur unintentionally due to the sound of the radio or audio equipment. Also, voice recognition may be difficult because the volume and frequency of voices vary depending on the passenger.

[0005] One example of a problem that the present invention aims to solve is improving the accuracy of voice recognition in a vehicle.

[0006] The invention described in claim 1 is a voice recognition device comprising: a voice input unit to which the voice of a passenger in a vehicle is input; a voice recognition unit that recognizes the voice input to the voice input unit as first voice information; a judgment unit that judges whether the similarity between the first voice information and pre-stored second voice information satisfies a predetermined criterion; a processing execution unit that executes a predetermined process if the similarity satisfies the predetermined criterion; and an adjustment unit that adjusts at least one of the similarity judgment rule and the predetermined criterion according to input from the passenger.

[0007] The invention described in claim 8 is a program for causing a computer realizing a voice recognition device to execute the following steps: inputting the voice of a passenger in a vehicle; recognizing the voice as first voice information; determining whether the similarity between the first voice information and pre-stored second voice information satisfies a predetermined criterion; executing a predetermined process if the similarity satisfies the predetermined criterion; and adjusting at least one of the similarity judgment rules and the predetermined criterion according to input from the passenger.

[0008] The invention described in claim 9 is a speech recognition method in which a computer realizing a speech recognition device inputs the speech of a passenger in a vehicle, recognizes the speech as first speech information, determines whether the similarity between the first speech information and pre-stored second speech information satisfies a predetermined criterion, executes a predetermined process if the similarity satisfies the predetermined criterion, and adjusts at least one of the similarity judgment rules and the predetermined criterion according to input from the passenger.

[0009] The invention described in claim 10 is a voice recognition system comprising an in-vehicle device mounted in a vehicle and a server, and further comprising: a voice input unit to which voices of passengers in the vehicle are input; a voice recognition unit that recognizes the voice input to the voice input unit as first voice information; a judgment unit that judges whether the similarity between the first voice information and pre-stored second voice information satisfies a predetermined criterion; a processing execution unit that executes a predetermined process if the similarity satisfies the predetermined criterion; and an adjustment unit that adjusts at least one of the similarity judgment rules and the predetermined criterion according to input from the passengers.

[0010] The above-mentioned objects, as well as other objects, features and advantages, will become more apparent from the preferred embodiments described below and the accompanying drawings.

[0011] 1 is a block diagram showing an outline of a speech recognition system and a speech recognition device according to a first embodiment; FIG. 2 is a simplified diagram for explaining a method by which a determination unit according to the first embodiment determines similarity; FIG. 3 is a simplified diagram for explaining a method by which a determination unit determines similarity when a plurality of pieces of second speech information are provided; FIG. 4 is a diagram showing an example of the hardware configuration of a speech recognition device; FIG. 5 is a flow diagram until the speech recognition device according to the first embodiment executes startup processing for a speech output unit; FIG. 6 is a block diagram showing an outline of a speech recognition system and a speech recognition device according to a second embodiment; FIG. 7 is a diagram showing an example of the data structure of second text information and information linked thereto stored in a storage unit according to the second embodiment; FIG. 8 is a simplified diagram for explaining a method by which a determination unit according to the second embodiment determines similarity; FIG. 9 is a simplified diagram for explaining a method by which a determination unit according to the second embodiment determines similarity when a plurality of pieces of second text information are provided; FIG. 10 is a diagram showing another example of the data structure of second text information and information linked thereto stored in a storage unit according to the second embodiment; FIG. 11 is a flow diagram until the speech recognition device according to the second embodiment executes startup processing for a speech output unit;

[0012] Hereinafter, embodiments of the present invention will be described with reference to the drawings. In all the drawings, like components are designated by like reference numerals, and the description thereof will be omitted as appropriate.

[0013] In the following description, each component of each device represents a functional block, not a hardware configuration. Each component of each device is realized by any combination of hardware and software, centered around the CPU of any computer, memory, a program loaded into the memory, a storage medium such as a hard disk that stores the program, and a network connection interface. There are many variations in the realization method and device.

[0014] 1 is a block diagram showing an outline of a speech recognition system 100 and a speech recognition device 1 according to a first embodiment. The speech recognition system 100 and the speech recognition device 1 will be described with reference to FIG.

[0015] (Speech Recognition System 100) The speech recognition system 100 includes an in-vehicle device 2 and a server 3. The in-vehicle device 2 is installed in a vehicle. The in-vehicle device 2 includes a voice output unit 2a, a microphone unit 2b, a camera unit 2c, and a server communication unit 2d. Although not shown, the in-vehicle device 2 may also include a display.

[0016] In the first embodiment, the voice output unit 2a is activated in response to a specific utterance from a passenger. The voice output unit 2a outputs a mechanical voice. For example, the voice output unit 2a outputs information about route guidance to a destination.

[0017] The microphone unit 2b receives voices uttered by passengers U (users). The passengers U include the driver and fellow passengers.

[0018] The camera unit 2c has an in-camera and an out-camera (not shown). The in-camera faces the interior of the vehicle and the driver's seat is included in the shooting range. The in-camera shoots the interior of the vehicle so that the driver can be seen. The out-camera faces outside the vehicle and shoots the situation outside the vehicle.

[0019] The server communication unit 2 d is configured to be able to communicate with the server 3 (= the in-vehicle device communication unit 70 ) via the network 101 .

[0020] In the first embodiment, the server 3 is provided outside the vehicle. The server 3 may be a so-called cloud server.

[0021] (Speech Recognition Device 1) In the first embodiment, the speech recognition device 1 is a server 3. The speech recognition device 1 includes a speech input unit 10, a speech recognition unit 20, a determination unit 30, a process execution unit 40, an adjustment unit 50, a storage unit 60, and an in-vehicle device communication unit 70.

[0022] (Audio Input Unit 10) The audio input unit 10 receives input of the audio of a passenger U in the vehicle. The audio input to the microphone unit 2b is transmitted to the audio input unit 10, whereby the audio of the passenger is input to the audio input unit 10.

[0023] (Speech recognition unit 20) The speech input to the speech input unit 10 is transmitted to the speech recognition unit 20. The speech recognition unit 20 recognizes the speech input to the speech input unit 10 as first speech information. The first speech information includes at least one of information on the speech waveform as a speech signal, information on the volume of the speech, and information on the frequency of the speech. In the first embodiment, the speech recognition unit 20 recognizes the first speech information using score information. The score information is information that quantifies the likelihood that the speech uttered by the passenger U is a predetermined term.

[0024] (Determination unit 30) The determination unit 30 compares the first speech information recognized by the speech recognition unit 20 with the second speech information, and determines whether the similarity between the two satisfies a predetermined standard. The second speech information is pre-stored in the storage unit 60. The second speech information includes at least one of information on a speech waveform for a specific phrase including the Japanese syllabary and prolonged vowels, information on the volume of the speech, and information on the frequency of the speech.

[0025] FIG. 2 is a simplified diagram illustrating a method for determining similarity by the determination unit 30 of the first embodiment. The determination unit 30 compares the first voice information with the second voice information to calculate score information. The score information indicates, for example, the similarity of the first voice information with respect to the second voice information. A threshold (i.e., a predetermined standard) is set for the score information, and when the score information of the first voice information of the voice uttered by the occupant U exceeds the threshold, the determination unit 30 determines that the first voice information and the second voice information are similar (or identical). In other words, when the score information of the first voice information exceeds the threshold, the determination unit 30 determines that the similarity between the first voice information and the second voice information satisfies the predetermined standard.

[0026] For example, a voice recognition method will be described for the case where the passenger U's speech "abc" is recognized. The voice recognition unit 20 recognizes the speech (=abc) spoken by the passenger U as first voice information (="xxx" (=C01 in FIG. 2), where "xxx" is voice waveform data, etc.). The determination unit 30 compares the first voice information with the second voice information "abc" (=C02 in FIG. 2) stored in the storage unit 60, and calculates score information (=C03 in FIG. 2).

[0027] The determination unit 30 compares the calculated score information with a threshold value (=C04 in FIG. 2) stored in the storage unit 60 to determine whether the threshold value has been exceeded. If the threshold value has been exceeded, the determination unit 30 determines that the first voice information and the second voice information are similar (or identical). In other words, if the threshold value has been exceeded, the determination unit 30 determines that the passenger U uttered "abc" (="Determination result: Similar" in FIG. 2). In the example of FIG. 2, the calculated score information is 100, which exceeds the threshold value of 80, so the determination unit 30 determines that the first voice information and the second voice information are similar.

[0028] 3 is a simplified diagram illustrating a method for the determination unit 30 to determine similarity when a plurality of pieces of second voice information are provided. In the first embodiment, a plurality of pieces of second voice information are stored in the storage unit 60. The words corresponding to the plurality of pieces of second voice information are different from each other as a whole but share parts in common. That is, one of the plurality of pieces of second voice information indicates the actual wording to be spoken by the passenger ("A-B-C" in FIG. 3). The remaining pieces of second voice information are either a part of the wording to be spoken by the passenger ("B-C" in FIG. 3) or a partially modified version of the wording to be spoken by the passenger ("Ebishi" in FIG. 3).

[0029] The determination unit 30 compares the first voice information with each of the plurality of pieces of second voice information and calculates score information for each of the plurality of pieces of second voice information. The determination unit 30 outputs a determination result for each of the plurality of pieces of second voice information. In the example of Fig. 3, the determination unit 30 determines that the first voice information and "B-C-" (second voice information) are similar.

[0030] In the first embodiment, when any of the score information related to each of the plurality of pieces of second voice information exceeds a threshold, the determination unit 30 may determine that the first voice information and the second voice information are similar. That is, when any of the score information related to each of the plurality of pieces of second voice information exceeds a threshold, the determination unit 30 may determine that the similarity between the first voice information and the second voice information satisfies a predetermined criterion.

[0031] 1 executes a predetermined process when the similarity between the first voice information and the second voice information satisfies a predetermined criterion. In the example of FIG. 2, the similarity between the first voice information and the second voice information satisfies the predetermined criterion (i.e., exceeds a threshold), so the process executing unit 40 executes the predetermined process.

[0032] In the first embodiment, the predetermined processing includes processing for starting up the voice output unit 2 a of the in-vehicle device 2. That is, in the first embodiment, when the passenger U utters a specific word and the word is recognized by the voice recognition device 1, the voice output unit 2 a is started up.

[0033] (Adjustment unit 50) The adjustment unit 50 shown in FIG. 1 adjusts at least one of the judgment rule for determining the similarity between the first voice information and the second voice information and the predetermined criterion according to input from the passenger U. In the first embodiment, the adjustment unit 50 adjusts at least one of the judgment rule and the criterion via a terminal managed by the passenger U. In the first embodiment, the adjustment unit 50 accepts input from a terminal managed (= owned) by the passenger U. In this case, the terminal managed by the passenger U is capable of communicating with at least one of the voice recognition device 1 (= server 3) and the in-vehicle device 2. The terminal is, for example, a smartphone, a tablet, or a PC (personal computer).

[0034] The steps in which the adjustment unit 50 adjusts at least one of the similarity determination rules and the predetermined criteria according to input from the passenger U will be described below.

[0035] When the adjustment unit 50 receives an input from the passenger U, the adjustment unit 50 adjusts at least one of the judgment rule and the predetermined criterion.

[0036] Specifically, for example, the passenger U may want to set a low voice recognition rate because unintentional voice recognition may occur due to radio, audio, or other external sounds. In such a case, the passenger U provides input so that the voice recognition rate of the voice recognition device 1 is lowered. The adjustment unit 50 then adjusts the threshold value to be higher (i.e., adjusts the predetermined standard). When the threshold value is higher, it becomes more difficult to determine that the first voice information and the second voice information are similar, making it more difficult to recognize the voice spoken by the passenger U. This lowers the voice recognition rate of the voice recognition device 1.

[0037] As another example, there are cases where the passenger U wishes to set a high voice recognition rate because voice recognition is difficult due to the characteristics of the passenger U's voice. In such a case, the passenger U provides input so as to increase the voice recognition rate of the voice recognition device 1. The adjustment unit 50 then adjusts the number of pieces of second voice information to be compared with the first voice information. In other words, based on the input from the passenger U, the adjustment unit 50 adjusts the judgment rule (i.e., the number of pieces of second voice information to be compared with the first voice information). By increasing the amount of second voice information to be compared with the first voice information, the voice recognition rate of the voice recognition device 1 improves.

[0038] 4 is a diagram showing an example of the hardware configuration of the speech recognition device 1. The speech recognition device 1 has a bus 1010, a processor 1020, a memory 1030, a storage device 1040, an input / output interface 1050, and a network interface 1060.

[0039] The bus 1010 is a data transmission path for transmitting and receiving data among the processor 1020, memory 1030, storage device 1040, input / output interface 1050, and network interface 1060. However, the method of connecting the processor 1020 and the like to each other is not limited to bus connection.

[0040] The processor 1020 is implemented by a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), or the like.

[0041] The memory 1030 is a main storage device realized by a RAM (Random Access Memory) or the like.

[0042] The storage device 1040 is an auxiliary storage device realized by removable media such as a hard disk drive (HDD), a solid state drive (SSD), or a memory card, or a read-only memory (ROM), and has a recording medium. The recording medium of the storage device 1040 stores program modules that realize each function of the voice recognition device 1 (e.g., the voice input unit 10, the voice recognition unit 20, the determination unit 30, the process execution unit 40, and the adjustment unit 50). The processor 1020 loads each of these program modules into the memory 1030 and executes them, thereby realizing each function corresponding to the program module. The storage device 1040 also functions as the memory unit 60.

[0043] The input / output interface 1050 is an interface for connecting the voice recognition device 1 to various input / output devices.

[0044] The network interface 1060 is an interface for connecting the speech recognition device 1 to a network. This network is, for example, a local area network (LAN) or a wide area network (WAN). The network interface 1060 may be connected to the network wirelessly or by wire. The speech recognition device 1 may communicate with the in-vehicle device 2 via the network interface 1060.

[0045] (Example of Operation of First Embodiment) Fig. 5 is a flow diagram showing the process until the voice recognition device 1 of the first embodiment executes the activation process of the voice output unit 2a. The flow until the voice recognition device 1 executes the activation process of the voice output unit 2a will be described with reference to Fig. 5.

[0046] First, in step S100, the voice of the passenger U is input to the voice input unit 10. Next, in step S110, the voice recognition unit 20 recognizes the voice input to the voice input unit 10 as first voice information. Next, in step S120, the determination unit 30 determines whether the similarity between the first voice information and pre-stored second voice information satisfies a predetermined criterion. If the predetermined criterion is satisfied (Yes in step S120), the process executes activation processing of the voice output unit 2a in step S130. If the predetermined criterion is not satisfied (No in step S120), the process returns to step S100 again. Note that if the predetermined criterion is not satisfied (No in step S120), the control may be terminated.

[0047] The voice recognition device 1 of this embodiment is characterized by comprising a judgment unit 30 that judges whether the similarity between the first voice information and the second voice information satisfies a predetermined criterion, a processing execution unit 40 that executes a predetermined process if the similarity satisfies the predetermined criterion, and an adjustment unit 50 that adjusts at least one of the similarity judgment rules and the predetermined criterion according to input from the passenger U.

[0048] Even if the system is controlled so that predetermined processing (such as startup processing of the voice output unit 2a, processing related to destination setting, and processing related to route guidance) is executed as a trigger for voice recognition of the voice (=keyword) spoken by the passenger U, the passenger U can adjust the voice recognition rate himself, thereby reducing the possibility of the predetermined processing being executed unintentionally and erroneously.

[0049] Furthermore, even if the voice uttered by the passenger U is difficult to recognize, the passenger U can increase the voice recognition rate of the voice recognition device 1 by himself.

[0050] Therefore, by providing a configuration with the above characteristics, the speech recognition accuracy of the speech recognition device 1 in a vehicle can be improved.

[0051] Furthermore, the adjustment unit 50 adjusts at least one of the judgment rules and the criteria via a terminal managed by the passenger U, so that the passenger U can easily adjust the voice recognition rate himself.

[0052] Furthermore, by configuring the process execution unit 40 to execute the startup process for the voice output unit 2a, it is possible to improve the accuracy of voice recognition of the wake word.

[0053] [Second embodiment] Fig. 6 is a block diagram showing an outline of a speech recognition system 100 and a speech recognition device 1 according to a second embodiment. Unlike the first embodiment, the speech recognition device 1 according to the second embodiment further includes a text conversion unit 80. In the first embodiment, first speech information is converted into first text information by the text conversion unit 80. The first text information includes character information such as the Japanese syllabary and prolonged sound symbols.

[0054] The second text information is stored in advance in the storage unit 60. The second text information includes character information such as the Japanese syllabary and prolonged sound symbols.

[0055] 7 is a diagram showing an example of the data structure of the second text information and information linked thereto stored in the storage unit 60 in the second embodiment. In the second embodiment, the second text information is linked with an ID, a priority, and information indicating whether or not it is a target for judgment. A priority is set for each of the multiple pieces of second voice information (= second text information).

[0056] 8 is a simplified diagram illustrating a method for determining similarity by the determination unit 30 of the second embodiment. The determination unit 30 determines whether the similarity between the first text information converted by the text conversion unit 80 and the second text information satisfies a predetermined criterion.

[0057] In the second embodiment, the determination unit 30 recognizes the proportion of the second text information contained in the first text information and determines whether or not the proportion satisfies a predetermined standard.

[0058] Assume that passenger U utters "A-B-C", the voice recognition unit 20 recognizes the utterance as first voice information, and the text conversion unit 80 converts it as "B-C" (= first text information).

[0059] The determination unit 30 compares the converted "B-C" with the "A-B-C" of the second text information to be determined, and calculates the percentage of the second text information contained in the first text information. In the case of Figure 8, the first text information and the second text information partially match in the "B-C" part, so the percentage is calculated to be approximately 67%.

[0060] For example, if the threshold value is set to 60% or more, the determination unit 30 determines that the first text information and the second text information are similar. In other words, the determination unit 30 determines that the similarity between the first text information and the second text information satisfies a predetermined standard.

[0061] FIG. 9 is a simplified diagram for explaining a method in which the determining unit 30 determines similarity when a plurality of pieces of second text information are provided.

[0062] In this case, assume that passenger U utters "A-B-C", the voice recognition unit 20 recognizes the utterance as the first voice information, and the text conversion unit 80 converts it as "K-B-C" (= first text information).

[0063] The second text information to be judged for similarity to the first text information may be set in advance. In Fig. 9, "A-B-C" and "K-B-C" are judged.

[0064] Since "ABC" in the second text information (ID: 001 in FIG. 9 ) and "KBC" in the first text information partially match with each other, the above-mentioned percentage is calculated as 67%. The determination unit 30 then compares the percentage value with a threshold value and determines whether it exceeds the threshold value. Since "ABC" in the second text information is below the threshold value of 90%, the determination unit 30 determines that the first text information and the second text information are dissimilar.

[0065] The "KBC" (ID: 003 in FIG. 9) in the second text information and the "KBC" in the first text information are a perfect match, and the above percentage is calculated as 100%. The determination unit 30 then compares the percentage value with a threshold value and determines whether it exceeds the threshold value. As for the "KBC" in the second text information, since it exceeds the threshold value of 90%, the determination unit 30 determines that the first text information and the second text information are similar (or identical).

[0066] In the second embodiment, when a plurality of pieces of second text information are provided, if the proportion of any of the second text information exceeds a threshold (if any of the second text information is similar to the first text information), the determination unit 30 determines that the similarity between the first text information and the second text information satisfies a predetermined criterion. In the example of Fig. 9, "A-B-C" does not exceed the threshold, but "K-B-C" does, so the determination unit 30 determines that the similarity satisfies the predetermined criterion.

[0067] In the second embodiment, similarly to the first embodiment, when an input is received from the passenger U, the adjustment unit 50 adjusts at least one of the judgment rule and the predetermined criterion.

[0068] In the second embodiment, the judgment rule includes judgment target information. The judgment target information is information for identifying, from among a plurality of pieces of second speech information (= second text information), second speech information (= second text information) to be used for the similarity judgment. The judgment target information also includes information on which second text information is to be compared with the first text information.

[0069] 9, the determination targets are "A-B-C" and "K-B-C." The determination target information includes, for example, information on which of the second text information IDs: 001 to 010 is to be set as the determination target.

[0070] In the second embodiment, the adjustment unit 50 may adjust the judgment target information in accordance with input from the occupant U. That is, the adjustment unit 50 may adjust which second text information is to be the judgment target in accordance with input from the occupant U. Furthermore, the adjustment unit 50 may preferentially include second text information with a relatively high priority in the judgment target.

[0071] The determination rule determines the number of pieces of second speech information (= second text information) used in comparison with the first speech information. The adjustment unit 50 adjusts the determination rule to increase or decrease the number of pieces of second speech information used in comparison with the first speech information. Furthermore, if priorities are set for the second speech information, when the adjustment unit 50 increases the number of pieces of second speech information, it may preferentially increase the second speech information with a higher priority, and when the adjustment unit 50 decreases the number of pieces of second speech information, it may preferentially decrease the second speech information with a higher priority. Note that when the adjustment unit 50 increases or decreases the number of pieces of second speech information, it is possible to arbitrarily select whether to increase or decrease the second speech information with a higher priority or the second speech information with a lower priority.

[0072] 10 is a diagram showing another example of the data structure of the second text information and information linked thereto stored in the storage unit 60 of the second embodiment. As shown in Fig. 9, the structure may be such that the information is collected in one data table T01, or as shown in Fig. 10, the structure may be such that multiple data tables (T01 to T10) are provided.

[0073] In the case of Fig. 9, the adjustment unit 50 may be configured to add or delete second text information and information linked thereto from the data table T01 in accordance with input from the passenger U. In the case of Fig. 10, the adjustment unit 50 may be configured to increase or decrease the number of data tables to be judged in accordance with input from the passenger U. In the case of Fig. 10, there are two data tables to be judged: T01 and T02.

[0074] (Example of Operation of Second Embodiment) FIG. 11 is a flow diagram showing the process up to when the speech recognition device 1 of the second embodiment executes the activation process of the speech output unit 2a. Unlike the speech recognition device 1 of the first embodiment, in step S111, the text conversion unit 80 converts the first speech information into first text information. In step S121, the determination unit 30 compares the first text information with the second text information and determines whether the similarity between the first text information and the second text information satisfies a predetermined criterion. If the predetermined criterion is satisfied (Yes in step S121), the activation process of the speech output unit 2a is executed in step S130. If the predetermined criterion is not satisfied (No in step S121), the process returns to step S100 again. Note that if the predetermined criterion is not satisfied (No in step S121), the control may be terminated.

[0075] In the second embodiment, as in the first embodiment, the speech recognition accuracy of the speech recognition device 1 in a vehicle can be improved. In the second embodiment, the determination unit 30 is configured to determine whether the similarity between the first text information and the second text information satisfies a predetermined criterion, thereby further improving the speech recognition accuracy of the speech recognition device 1.

[0076] The judgment unit 30 recognizes the proportion of the second text information contained in the first text information and determines whether it satisfies a predetermined standard, thereby making it possible to clearly judge the similarity, thereby further improving the speech recognition accuracy of the speech recognition device 1.

[0077] By the adjustment unit 50 adjusting the judgment target information in accordance with the input from the passenger U, the speech recognition accuracy of the speech recognition device 1 can be effectively improved.

[0078] The adjustment unit 50 preferentially includes the second speech information having a relatively high priority in the determination target, thereby more effectively improving the speech recognition accuracy of the speech recognition device 1 .

[0079] Although the embodiments have been described above with reference to the drawings, these are merely examples of the present invention, and various other configurations can also be adopted.

[0080] In the above description, the voice recognition device 1 has been described as the server 3, but the voice recognition device 1 may be an in-vehicle device 2. In this case, the voice input unit 10, the voice recognition unit 20, the determination unit 30, the process execution unit 40, and the adjustment unit 50 may be included in the in-vehicle device 2.

[0081] Furthermore, it may be arbitrarily selected whether the voice input unit 10, the voice recognition unit 20, the determination unit 30, the process execution unit 40, and the adjustment unit 50 are mounted on the in-vehicle device 2 or the server 3.

[0082] The storage unit 60 may be provided outside the voice recognition device 1. The predetermined processing may include processing related to setting a destination for the vehicle and processing related to route guidance, in addition to the startup processing of the voice output unit 2a.

[0083] In addition, although the flowcharts used in the above description show multiple steps (processes) in a sequential order, the order of steps executed in each embodiment is not limited to the order shown. In each embodiment, the order of the steps shown in the drawings can be changed as long as it does not cause any problems in terms of content. Furthermore, the above-described embodiments can be combined as long as the content is not contradictory.

[0084] Examples of reference forms are given below. 1. A voice recognition device comprising: a voice input unit to which voice of a passenger in a vehicle is input; a voice recognition unit that recognizes the voice input to the voice input unit as first voice information; a determination unit that determines whether similarity between the first voice information and pre-stored second voice information satisfies a predetermined criterion; a process execution unit that executes a predetermined process if the similarity satisfies the predetermined criterion; and an adjustment unit that adjusts at least one of the similarity determination rule and the predetermined criterion in accordance with input from the passenger. 2. The voice recognition device described in 1., in which the first voice information is converted into first text information, and the determination unit determines whether the similarity between the first text information and pre-stored second text information satisfies the predetermined criterion. 3. The voice recognition device described in 2., in which the determination unit recognizes the proportion of the second text information included in the first text information and determines whether the predetermined criterion is satisfied. 4. The voice recognition device described in 1. to 3. 5. A speech recognition device described in any one of above, wherein the judgment rule includes judgment target information for identifying second speech information to be the target of the similarity judgment from among the plurality of second speech information, and the adjustment unit adjusts the judgment target information in accordance with input from the occupant. 5. A speech recognition device described in above 4, wherein a priority is set for each of the plurality of second speech information, and the adjustment unit preferentially includes second speech information with a relatively high priority in the judgment target. 6. A speech recognition device described in any one of above 1 to 5, wherein the adjustment unit adjusts at least one of the judgment rule and the criterion via a terminal managed by the occupant. 7. A speech recognition device described in any one of above 1 to 6, wherein the predetermined processing includes a process of starting up a speech output unit.8. A program causing a computer implementing a speech recognition device to execute the following steps: inputting the speech of a passenger in a vehicle, recognizing the speech as first speech information, determining whether the similarity between the first speech information and pre-stored second speech information satisfies a predetermined criterion, executing a predetermined process if the similarity satisfies the predetermined criterion, and adjusting at least one of the similarity judgment rule and the predetermined criterion according to input from the passenger. 9. A speech recognition method in which a computer implementing a speech recognition device inputs the speech of a passenger in a vehicle, recognizes the speech as first speech information, determines whether the similarity between the first speech information and pre-stored second speech information satisfies a predetermined criterion, executes a predetermined process if the similarity satisfies the predetermined criterion, and adjusts at least one of the similarity judgment rule and the predetermined criterion according to input from the passenger. 10. A voice recognition system comprising an in-vehicle device mounted in a vehicle and a server, further comprising: a voice input unit to which voices of passengers in the vehicle are input; a voice recognition unit that recognizes the voice input to the voice input unit as first voice information; a judgment unit that judges whether a similarity between the first voice information and pre-stored second voice information satisfies a predetermined criterion; a processing execution unit that executes a predetermined process if the similarity satisfies the predetermined criterion; and an adjustment unit that adjusts at least one of the similarity judgment rule and the predetermined criterion according to input from the passenger.

[0085] REFERENCE SIGNS LIST 1 Speech recognition device 2 In-vehicle device 2a Speech output unit 3 Server 10 Speech input unit 20 Speech recognition unit 30 Determination unit 40 Processing execution unit 50 Adjustment unit 60 Storage unit 70 In-vehicle communication unit 100 Speech recognition system

Claims

1. a voice input unit to which voices of passengers in the vehicle are input; a voice recognition unit that recognizes the voice input to the voice input unit as first voice information; a determination unit that determines whether a similarity between the first voice information and a plurality of second voice information pieces that are prestored and indicate a plurality of words that are different from each other and similar to each other by being partially common to each other satisfies a predetermined criterion; a processing execution unit that executes a predetermined process when the similarity satisfies the predetermined criterion; and an adjustment unit that adjusts at least one of the similarity judgment rule and the predetermined criterion according to an input from the passenger.

2. 2. The speech recognition device according to claim 1, The first audio information is converted into first text information; The determination unit determines whether the similarity between the first text information and pre-stored second text information satisfies the predetermined criterion.

3. 3. The speech recognition device according to claim 2, The determination unit recognizes a ratio of the second text information contained in the first text information and determines whether the predetermined criterion is satisfied.

4. In the speech recognition device according to claim 2 or 3, A voice recognition device wherein the adjustment unit changes the number of pieces of second text information or the wording set as the second text information when adjusting the judgment rule in accordance with input from the occupant.

5. 5. The speech recognition device according to claim 4, A priority is set for each of the plurality of second voice information, The adjustment unit preferentially includes second speech information having a relatively high priority in a determination target.

6. 4. The speech recognition device according to claim 1, The adjustment unit adjusts at least one of the judgment rule and the criterion via a terminal managed by the passenger.

7. 4. The speech recognition device according to claim 1, The predetermined process includes a process of starting up a voice output unit.

8. A computer that realizes a speech recognition device A procedure for inputting the voice of a passenger in a vehicle; recognizing the speech as first speech information; A step of determining whether or not a similarity between the first voice information and a plurality of second voice information pieces, which are pre-stored voice information pieces and indicate a plurality of words that are different from each other and similar to each other by being partially common to each other, satisfies a predetermined criterion; a step of executing a predetermined process if the similarity satisfies the predetermined criterion; A program for executing a procedure for adjusting at least one of the similarity judgment rule and the predetermined criteria according to input from the passenger.

9. A computer that realizes a speech recognition device, Input the voice of the passengers in the vehicle, Recognizing the speech as first speech information; determining whether or not a similarity between the first voice information and a plurality of second voice information pieces, which are pre-stored voice information pieces and indicate a plurality of words that are different from each other and similar to each other by having partial commonality with each other, satisfies a predetermined criterion; Executing a predetermined process if the similarity satisfies the predetermined criterion; The method of claim 1, further comprising adjusting at least one of the similarity determination rules and the predetermined criteria according to an input from the passenger.

10. A voice recognition system including an in-vehicle device installed in a vehicle and a server, a voice input unit to which voices of passengers in the vehicle are input; a voice recognition unit that recognizes the voice input to the voice input unit as first voice information; a determination unit that determines whether a similarity between the first voice information and a plurality of second voice information pieces that are prestored and indicate a plurality of words that are different from each other and similar to each other by being partially common to each other satisfies a predetermined criterion; a processing execution unit that executes a predetermined process when the similarity satisfies the predetermined criterion; The speech recognition system further comprises an adjustment unit that adjusts at least one of the similarity judgment rule and the predetermined criterion according to an input from the passenger.