JP2007219207A - Speech recognition device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a speech recognition device capable of appropriately performing a speech recognition even when a person other than a speech inputting object person is present under its use environment. <P>SOLUTION: The speech recognition device which recognizes a speech that the speech inputting object person (driver) speaks is equipped with a first decision means of deciding whether there is the speech spoken by the driver, a second decision means of deciding whether there is a speech that the person (passenger) other than the speech inputting object person speaks, and a condition meeting judging means of judging whether a speech recognition start condition of speech recognition is met, the speech recognition start condition including driver's speaking and no speaking of the passenger right after the driver's speaking. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a speech recognition apparatus, and more particularly, to a speech recognition apparatus that employs speech recognition technology and recognizes speech emitted from a user.

  Voice recognition devices are used in various fields, and are used in in-vehicle devices such as navigation systems. The voice emitted from the driver is input by a microphone, and the input voice is converted into an electric signal by the microphone. The input voice converted into the electrical signal is analyzed by the voice processing unit, the command uttered by the driver is recognized, and the navigation device operates according to the recognition result.

In a speech recognition apparatus, since a high speech recognition rate is very important, it is also important to specify a processing section that is a period for performing recognition processing on speech input by a microphone. If the recognition process is always performed without specifying the processing section, malfunction may occur due to the influence of conversation with passengers, music from the car stereo, noise, and the like.
For this reason, many conventional speech recognition devices are provided with an utterance switch (talk switch) that is operated by a user when inputting voice. The utterance switch includes an utterance start switch and a press talk switch.

The utterance start switch is operated immediately before the start of utterance by the user. When the utterance start switch is operated, a recognition process is performed on the voice input from the microphone immediately after the operation.
On the other hand, the press talk switch is continuously pressed from the start to the end of the utterance by the user, and while the press talk switch is being pressed, the recognition process for the voice input by the microphone is performed.

  However, in such a speech recognition apparatus, the user must operate the speech switch every time he speaks, and there is a problem that the operation becomes very complicated. In particular, the operation of the speech switch by the driver while traveling is not preferable. If a voice recognition device is employed, it is desirable that manual operation is unnecessary.

  As a technique for solving such a problem, for example, in Patent Document 1 below, when the user's face is facing the microphone, the user's lips move, or the user's line of sight looks at the microphone. A technique is disclosed in which voice recognition is started when a user's utterance is determined by detecting such an appearance state and when it is determined that the utterance by the user has started.

However, when there is another person other than the voice input target person in the usage environment, such as when there is a passenger in the vehicle, the person is talking to another person or heading to the microphone for voice input. The voice recognition process is started and may cause malfunction even though it is not necessary if necessary. For example, it is conceivable that the voice recognition process is started in a conversation with a passenger.
Japanese Patent Laid-Open No. 11-352987

Means for solving the problems and their effects

  The present invention has been made in view of the above problems, and provides a speech recognition device capable of realizing appropriate speech recognition even in the presence of another person other than the speech input target person. The purpose is to do.

  In order to achieve the above object, a speech recognition device (1) according to the present invention is a speech recognition device that recognizes speech uttered by a speech input target person, and determines whether or not the speech input target person utters. A determination unit; a second determination unit that determines the presence or absence of utterance by a person other than the voice input target person; and a condition establishment determination unit that determines whether or not a voice recognition start condition for starting voice recognition is satisfied. Along with the voice recognition start condition, it is determined that there is utterance by the voice input target person in the first determination means, and the second determination means until a predetermined period has elapsed from the utterance by the voice input target person. Are not determined to be uttered by a person other than the voice input target person.

  According to the voice recognition device (1), the voice recognition start condition for starting voice recognition includes that the voice input target person (for example, the driver) speaks. Speech recognition can be started as a trigger. Therefore, voice recognition can be started at a timing desired by the voice input target person without causing the voice input target person to perform a manual operation such as pressing a switch.

  Moreover, a person other than the voice input target person (for example, a fellow passenger) speaks until the predetermined period has passed since the voice input target person uttered (for example, until 2 seconds have elapsed from the end of the utterance). Therefore, if a person other than the voice input target person speaks immediately after the voice input target person speaks, the voice recognition is not started.

  When a person other than the voice input target person speaks immediately after the voice input target person speaks, there is a high possibility that a conversation is being exchanged between them. Therefore, when there is a high possibility that the utterance by the voice input target person is a part of the conversation with a person other than the voice input target person, the voice recognition is not started. Can be prevented.

For example, in the following cases, voice recognition is not started.
1. Speech by the driver who is the target of voice input.
2. Speech produced by a passenger who is not a voice input subject within 2 seconds from the speech of 1 above.
3. The utterance by the driver after the utterance of 2 above.
4). Speaking by passengers within 2 seconds of utterance 3 above.

  Further, in the voice recognition device (2) according to the present invention, in the voice recognition device (1), the utterer is the voice input target person based on the personality information included in the voice input by the voice input means. A voicing main judgment means for judging whether or not, and based on the judgment result by the utterance main judgment means, the judgment by the first judgment means and the judgment by the second judgment means are made. It is characterized by being.

  According to the voice recognition device (2), it is determined whether or not the utterer is the voice input target person based on the personality information included in the voice input by the voice input means, and based on the determination result. Thus, the presence or absence of utterance by the voice input target person and the presence or absence of utterance by a person other than the voice input target person are determined. Examples of the personality information include a voice print, formant (resonance frequency of the vocal tract), and the like.

  If it is determined that the utterer is a voice input target person (or the possibility is high), it is determined that the voice input target person uttered, while the utterance person is not the voice input target person (or the possibility). If it is determined that a person other than the voice input target person has uttered, it is determined. Therefore, if there is personality information included in the voice of the voice input target person, these determinations can be made appropriately. For example, if there is voiceprint data of a driver who is a voice input target, it is possible to appropriately determine the voice of the driver and the voice of the passenger.

  Further, in the voice recognition device (3) according to the present invention, in the voice recognition device (1), based on the sound source direction obtained from the voice input by the voice input means, is the utterer the voice input target person? A voicing main judgment means for judging whether or not, and based on a judgment result by the utterance main judgment means, the judgment by the first judgment means and the judgment by the second judgment means are made. It is characterized by that.

  According to the voice recognition device (3), it is determined whether or not the utterer is a voice input target person based on the sound source direction obtained from the voice input by the voice input means, and based on the determination result. The presence or absence of utterance by the voice input target person and the presence or absence of utterance by a person other than the voice input target person are determined.

  If it is determined that the utterer is a voice input target person (or the possibility is high), it is determined that the voice input target person uttered, while the utterance person is not the voice input target person (or the possibility). If it is determined that a person other than the voice input target person has uttered, it is determined. Therefore, if there is data indicating where the voice input target person exists, these determinations can be made appropriately. For example, if the driver is a voice input target, if there is data indicating the position of the driver's seat, the driver's voice and the passenger's voice can be appropriately determined.

  Further, the speech recognition device (4) according to the present invention is a speech input target person obtained from an image input by the image input means in the speech recognition device (1), a person other than the speech input target person, or a speech. Based on the appearance state of a person other than the input target person and the voice input target person, a voice main judgment means for judging whether or not the voice main person is the voice input target person is provided, and based on the judgment result by the voice main judgment means The determination by the first determination means and the determination by the second determination means are performed.

  According to the voice recognition device (4), the voice input target person obtained from the voice input by the image input means, the person other than the voice input target person, or the voice input target person and the person other than the voice input target person. Based on the appearance state, it is determined whether or not the utterer is a voice input target person. Based on the determination result, the presence or absence of utterance by the voice input target person and the utterance by a person other than the voice input target person Is determined.

  In the appearance state where the face faces the microphone or the lips move, the person is likely to be the main speaker. For example, if the lip of the voice input target person is moving when the utterance is detected with the microphone, the utterer can be determined as the voice input target person, while the voice input is performed when the utterance is detected with the microphone. If the lips of a person other than the target person are moving, the utterer can be determined to be a person other than the voice input target person. Further, when the utterance of the voice input target person is not moving when the utterance is detected by the microphone, the utterer can be determined to be a person other than the voice input target person.

  If it is determined that the utterer is a voice input target person (or the possibility is high), it is determined that the voice input target person uttered, while the utterance person is not the voice input target person (or the possibility). If it is determined that a person other than the voice input target person has uttered, it is determined. Therefore, these determinations can be made appropriately by monitoring the appearance of the voice input target person, the person other than the voice input target person, or the voice input target person and the person other than the voice input target person. For example, if the face of the driver who is the voice input target is monitored, the utterance of the driver and the utterance of the passenger can be appropriately determined.

  In addition, the speech recognition device (5) according to the present invention may be configured such that, in any of the speech recognition devices (1) to (4), the second period until the predetermined period elapses after the speech is input by the speech input target person. When it is determined by the determination means that a person other than the voice input target person has uttered, there is provided a holding period setting means for setting a holding period in which the voice recognition start condition is not satisfied.

  Even if there was no utterance by a person other than the voice input target until the predetermined period has passed since the voice by the voice input target person, if there was a conversation between the two until then, the voice input at that time The utterance by the target person is not for the operation target device but is likely to be part of the conversation.

For example, the following cases can be considered.
1. Speech by the driver who is the target of voice input.
2. Speech produced by a passenger who is not a voice input subject within 2 seconds from the speech of 1 above.
3. The utterance by the driver within 2 seconds from the above utterance 2.
4). Even if 2 seconds have passed since the above utterance 3, there is no utterance by the passenger.
The utterance by the driver in the above 3 is not for the operation target device but is likely to be part of the conversation.

  According to the voice recognition device (5), when it is determined that a person other than the voice input target person has uttered before the predetermined period has elapsed since the voice input target person uttered (that is, a conversation between the two is performed). When there is a high possibility that the voice recognition is exchanged, a holding period (for example, 10 seconds) in which the voice recognition start condition is not satisfied is set. Thereby, it is possible to prevent the voice recognition from being started by the utterance by the driver in the above 3, and to prevent unnecessary voice recognition from starting.

  Embodiments of a speech recognition apparatus according to the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram schematically showing a main part of a navigation system in which the speech recognition apparatus according to the embodiment (1) is employed. In the figure, reference numeral 1 denotes a speech recognition device. The speech recognition device 1 includes an A / D converter 2 that converts a speech signal from the microphone 6 into a digital signal, and a digital signal that is obtained from the microphone 6 (about several seconds). FIFO (first-in first-out) type buffer memory 3 for storing voice signals after conversion, EEPROM 4 in which personality information (for example, voiceprint data) included in the voice of the driver who is the voice input target is stored, and CPU And a voice processing unit 5 having a ROM, a RAM, and the like.

The voice processing unit 5 determines whether or not the speaker is the driver (speech input target person) based on the personality information included in the voice input by the microphone 6 and the personality information stored in the EEPROM 4. And the like (speaker recognition function) and a function of performing voice recognition processing on a voice signal from the microphone 6 are provided. In addition, a signal corresponding to a voice command obtained by the recognition processing in the voice processing unit 5 is transmitted to the navigation device 7 by in-vehicle communication.
As shown in FIG. 2, the microphone 6 is installed at a substantially central portion in front of the driver seat 9 and the passenger seat 10 of the vehicle 8, and appropriately reproduces both the voice emitted by the driver and the voice emitted by the passenger. You can get it.

  Next, the processing operation [1] performed by the speech processing unit 5 in the speech recognition apparatus 1 according to Embodiment (1) will be described based on the flowchart shown in FIG. This processing operation [1] is an operation performed in response to an activation request from the navigation device 7 or an ignition switch ON.

  First, it is determined whether or not a sound is input from the microphone 6 (for example, whether or not a sound having a volume of a certain level or more is input) (step S1). If it is determined that a sound is input, it is included in the input sound. Speaker recognition processing is performed based on the personality information to be recorded and the personality information stored in the EEPROM 4 (included in the driver's voice) (step S2). It is determined whether or not there is (step S3). On the other hand, if it is determined that there is no voice input, the process directly returns to step S1.

  If it is determined in step S3 that the speaker is the driver, that is, if it is determined that one of the voice recognition start conditions is satisfied, next, whether or not the voice input with the microphone 6 has ended (for example, there is It is determined whether or not there is no voice input with a volume higher than the volume (step S4). On the other hand, if it is determined that the speaker is not the driver, the process directly returns to step S1.

If it is determined in step S4 that the voice input has been completed (that is, the utterance by the driver has been completed), it is next determined whether there has been a voice input with the microphone 6 (that is, whether there has been another voice input). ) Is determined (step S5).
If it is determined that there is a voice input, speaker recognition processing is performed based on the personality information included in the input voice and the personality information stored in the EEPROM 4 (included in the driver's voice) (step S6). ), It is determined whether or not the speaker is a passenger (a person other than the voice input target person) (step S7).

If it is determined that the speaker is not a passenger, it is then determined whether or not 2 seconds have elapsed since the driver's end of speaking (step S8). If it is determined in step S5 that there is no voice input, the process proceeds to step S8 and the same determination process as described above is performed. Here, it is determined that 2 seconds have elapsed from the end of the driver's utterance when the passenger does not utter until 2 seconds have elapsed from the end of the driver's utterance. That is, it is a case where the passenger does not speak immediately after the driver speaks.
When the passenger speaks immediately after the driver speaks, there is a high possibility that a conversation is being exchanged between the two. In other words, there is a high possibility that the driver's utterance without the utterance by the passenger immediately after that is not a conversation but a voice operation intention display for the navigation device 7.

  In step S8, if it is determined that 2 seconds have elapsed since the end of the driver's utterance, that is, if it is determined that one of the voice recognition start conditions is satisfied, then the voice signal is read from the buffer memory 3, A recognition process is performed on the voice signal (step S9), and a signal corresponding to the voice command obtained by the recognition process is transmitted to the navigation device 7 by in-vehicle communication (step S10).

  Thereafter, it is determined whether or not the voice recognition end condition is satisfied (step S11). If it is determined that the voice recognition end condition is satisfied, the process returns to step S1, and it is determined that the voice recognition end condition is not satisfied. If it does, it will return to step S9 and will continue a speech recognition process. Note that the voice recognition end condition includes, for example, that voice input is not continuously detected for a certain period of time. On the other hand, if it is determined in step S8 that 2 seconds have not elapsed since the end of the driver's speech, the process returns to step S5.

  In step S7, if it is determined that the speaker is the passenger, that is, if it is determined that the passenger has spoken within 2 seconds after the end of the driver's speech, the driver's speech is Therefore, it is determined that the voice recognition start condition is not satisfied, and the process returns to step S1.

  According to the voice recognition device according to the above embodiment (1), since the voice recognition start condition for starting voice recognition includes that the voice input target person (driver) uttered, the voice input target person Voice recognition can be started with the utterance of. Therefore, voice recognition can be started at a timing desired by the voice input target person without causing the voice input target person to perform a manual operation such as pressing a switch.

  In addition, since the voice recognition start condition includes that a person other than the voice input target person (passenger) does not utter until two seconds have elapsed from the end of the utterance by the voice input target person. If a person other than the voice input target person utters immediately after the person utters, the speech recognition is not started.

  When a person other than the voice input target person speaks immediately after the voice input target person speaks, there is a high possibility that a conversation is being exchanged between them. Therefore, when there is a high possibility that the utterance by the voice input target person is part of a conversation with a person other than the voice input target person, the voice recognition is not started. Can be prevented.

  FIG. 4 is a block diagram schematically showing a main part of a navigation system in which the speech recognition apparatus according to the embodiment (2) is employed. In the figure, reference numeral 21 denotes a speech recognition device. The speech recognition device 21 is an A / D converter 22 and 23 that converts speech signals from the microphones 26 and 27 into digital signals, and a few seconds obtained from the microphones 26 and 27. It comprises a FIFO type buffer memory 24 for storing a sound signal (converted into a digital signal) and a sound processing unit 25 having a CPU, ROM, RAM and the like.

  The voice processing unit 25 is equipped with a function of specifying a sound source direction from voices input from the microphones 26 and 27, a function of performing voice recognition processing on voice signals from the microphones 26 and 27, and the like. In addition, a signal corresponding to the voice command obtained by the recognition processing in the voice processing unit 25 is transmitted to the navigation device 7 by in-vehicle communication.

  The microphones 26 and 27 have directivity. As shown in FIG. 5, the microphone 26 is installed in front of the passenger seat 10 of the vehicle 8 so that the directivity faces the driver seat 9. It is installed in front of the driver's seat 9 of the vehicle 8 so that its directivity faces the passenger seat 10, and the sound emitted by the driver is appropriately acquired by the microphone 26 and the sound emitted by the passenger is appropriately acquired by the microphone 27. Be able to.

  Next, the processing operation [2] performed by the voice processing unit 25 in the voice recognition device 21 according to Embodiment (2) will be described based on the flowchart shown in FIG. This processing operation [2] is an operation performed in response to an activation request from the navigation device 7 or an ignition switch ON.

  First, it is determined whether or not a sound is input from the microphones 26 and 27 (for example, whether or not a sound input having a volume higher than a certain level is input) (step S21). , 27 is performed to identify the sound source direction from the voice input (step S22), and it is determined whether or not the speaker is the driver of the voice input target person (step S23). On the other hand, if it is determined that there is no voice input, the process directly returns to step S21.

  When the volume input to the microphone 26 is larger than the volume input to the microphone 27, it can be determined that the sound source direction is the direction of the driver's seat 9, and vice versa. When the volume is smaller than the volume input to the microphone 27, it can be determined that the sound source direction is the direction of the passenger seat 10.

  If it is determined in step S23 that the speaker is the driver, that is, if it is determined that one of the voice recognition start conditions is satisfied, then whether or not the voice input with the microphones 26 and 27 has ended (for example, Then, it is determined whether or not there is no more voice input with a volume higher than a certain level (step S24). On the other hand, if it is determined that the speaker is not the driver, the process directly returns to step S21.

If it is determined in step S24 that the voice input has been completed (that is, the utterance by the driver has been completed), next, whether or not there has been a voice input by the microphones 26 and 27 (that is, another voice input has been made again) (No) is determined (step S25).
If it is determined that there is a voice input, a process of specifying the sound source direction from the voices input from the microphones 26 and 27 is performed (step S26), and whether the speaker is a passenger (a person other than the voice input target person). It is determined whether or not (step S27).

If it is determined that the speaker is not a passenger, it is then determined whether 2 seconds have elapsed since the driver's utterance was completed (step S28). If it is determined in step S25 that there is no voice input, the process proceeds to step S28, and the same determination process as described above is performed. Here, it is determined that 2 seconds have elapsed from the end of the driver's utterance when the passenger does not utter until 2 seconds have elapsed from the end of the driver's utterance. That is, it is a case where the passenger does not speak immediately after the driver speaks.
When the passenger speaks immediately after the driver speaks, there is a high possibility that a conversation is being exchanged between the two. In other words, there is a high possibility that the driver's utterance without the utterance by the passenger immediately after that is not a conversation but a voice operation intention display for the navigation device 7.

  In step S28, if it is determined that two seconds have elapsed since the end of the driver's utterance, that is, if it is determined that one of the voice recognition start conditions is satisfied, then the voice signal is read from the buffer memory 24, A recognition process is performed on the voice signal (step S29), and a signal corresponding to the voice command obtained by the recognition process is transmitted to the navigation device 7 by in-vehicle communication (step S30).

  Thereafter, it is determined whether or not the voice recognition end condition is satisfied (step S31). If it is determined that the voice recognition end condition is satisfied, the process returns to step S21, and it is determined that the voice recognition end condition is not satisfied. If it does, it will return to step S29 and will continue a speech recognition process. Note that the voice recognition end condition includes, for example, that voice input is not continuously detected for a certain period of time. On the other hand, if it is determined in step S28 that 2 seconds have not elapsed since the end of the driver's speech, the process returns to step S25.

  In step S27, if it is determined that the speaker is the passenger, that is, if it is determined that the passenger has spoken within 2 seconds from the end of the driver's speech, the driver's speech is between the two. Therefore, since the voice recognition start condition is not satisfied, the process returns to step S21.

  FIG. 7 is a block diagram schematically showing a main part of a navigation system in which the speech recognition apparatus according to the embodiment (3) is employed. In the figure, reference numeral 31 denotes a speech recognition device. The speech recognition device 31 has an A / D converter 32 that converts a speech signal from the microphone 36 into a digital signal, and a digital signal that is obtained from the microphone 36 (about several seconds). FIFO type buffer memory 33 for storing audio signals (after conversion), audio processing unit 34 having CPU, ROM, RAM, etc., CPU, ROM, RAM, etc., and receiving image data from CCD camera 37 And an image processing unit 35 to be processed.

  The voice processing unit 34 is equipped with a function for performing voice recognition processing on a voice signal from the microphone 36. In addition, a signal corresponding to the voice command obtained by the recognition processing by the voice processing unit 34 is transmitted to the navigation device 7 by in-vehicle communication. The image processing unit 35 monitors the appearance of the driver and passengers (especially the movement of the lips) based on the image data obtained from the CCD camera 37, and the speaker is the driver or passenger who is the voice input subject. It is equipped with a function to identify which one is.

  The microphone 36 has directivity, and as shown in FIG. 8, the microphone 36 is installed in front of the driver seat 9 of the vehicle 8 so that the directivity faces the driver seat 9. You can get it properly. The CCD camera 37 is installed in the front center part of the driver's seat 9 and the passenger seat 10 so that the appearance of the driver and passengers can be photographed.

  Next, the processing operation [3] performed by the voice processing unit 34 in the voice recognition device 31 according to Embodiment (3) will be described based on the flowchart shown in FIG. This processing operation [3] is an operation performed in response to an activation request from the navigation device 7 or an ignition switch ON.

  First, it is determined whether or not a sound is input from the microphone 36 (for example, whether or not a sound having a volume higher than a certain level is input) (step S41). If it is determined that a sound is input, the image processing unit 35 is determined. Is requested to transmit the utterance person specifying information (step S42), and based on the utterance person specifying information sent from the image processing unit 35, it is determined whether or not the uttering person is the driver of the voice input target person. (Step S43). On the other hand, if it is determined that there is no voice input, the process directly returns to step S41.

  If it is determined in step S43 that the speaker is the driver, that is, if it is determined that one of the voice recognition start conditions is satisfied, then whether or not the voice input with the microphone 36 has ended (for example, there is It is determined whether or not there is no voice input with a volume higher than the volume (step S44). On the other hand, if it is determined that the speaker is not the driver, the process directly returns to step S41.

If it is determined in step S44 that the voice input has ended (that is, the utterance by the driver has ended), next, whether or not there has been a voice input with the microphone 36 (that is, whether another voice input has been made again). ) Is determined (step S45).
If it is determined that there has been a voice input, the image processing unit 35 is requested to transmit the speaker main specifying information (step S46), and based on the speaker main specifying information sent from the image processing unit 35, the speaker speaks with the passenger. It is determined whether or not it is a person other than the voice input target person (step S47).

If it is determined that the speaker is not a fellow passenger, it is then determined whether or not 2 seconds have elapsed since the end of the driver's utterance (step S48). If it is determined in step S45 that there is no voice input, the process proceeds to step S48 and the same determination process as described above is performed. Here, it is determined that 2 seconds have elapsed from the end of the driver's utterance when the passenger does not utter until 2 seconds have elapsed from the end of the driver's utterance. That is, it is a case where the passenger does not speak immediately after the driver speaks.
When the passenger speaks immediately after the driver speaks, there is a high possibility that a conversation is being exchanged between the two. In other words, there is a high possibility that the driver's utterance without the utterance by the passenger immediately after that is not a conversation but a voice operation intention display for the navigation device 7.

  If it is determined in step S48 that two seconds have elapsed since the end of the driver's utterance, that is, if it is determined that one of the voice recognition start conditions is satisfied, the voice signal is then read from the buffer memory 33, A recognition process is performed on the voice signal (step S49), and a signal corresponding to the voice command obtained by the recognition process is transmitted to the navigation device 7 by in-vehicle communication (step S50).

  Thereafter, it is determined whether or not the voice recognition end condition is satisfied (step S51). If it is determined that the voice recognition end condition is satisfied, the process returns to step S41, and it is determined that the voice recognition end condition is not satisfied. Then, the process returns to step S49 and the voice recognition process is continued. Note that the voice recognition end condition includes, for example, that voice input is not continuously detected for a certain period of time. On the other hand, if it is determined in step S48 that 2 seconds have not elapsed since the end of the driver's speech, the process returns to step S45.

  In step S47, if it is determined that the speaker is the passenger, that is, if it is determined that the passenger has spoken within 2 seconds from the end of the driver's speech, the driver's speech is Therefore, since the voice recognition start condition is not satisfied, the process returns to step S41.

  In the speech recognition apparatuses according to the above embodiments (1) to (3), if there is a utterance by the passenger immediately after the utterance of the driver who is the voice input target (“Y” in steps S7, S27, and S47) Judgment), the voice recognition start condition is not established, and the process returns to steps S1, S21, and S41. However, in the voice recognition device according to another embodiment, for example, as shown in FIG. If there is an utterance by the passenger immediately after the driver's utterance, a suspension period (for example, 10 seconds) may be provided in which the process proceeds to step S7A and is suspended. The reason for providing such a holding period is to prevent the start of speech recognition processing by conversation in the following cases.

Even if there is no utterance by the passenger immediately after the utterance by the driver, the utterance by the driver at that time is not for the navigation device 7 as long as there is a conversation between them. This is likely to be a response to the person. For example, the following cases can be considered.
1. Speaking by the driver.
2. Speaking by passengers within 2 seconds of utterance 1 above.
3. The utterance by the driver within 2 seconds from the above utterance 2.
4). Even if 2 seconds have passed since the above utterance 3, there is no utterance by the passenger.
It is highly possible that the utterance by the driver in the above 3 is not a response to the navigation device 7 but a response to the passenger.

It is the block diagram which showed roughly the principal part of the navigation system by which the speech recognition apparatus which concerns on embodiment (1) of this invention was employ | adopted. It is explanatory drawing for demonstrating the installation place of a microphone. It is the flowchart which showed the processing operation which the speech processing part in the speech recognition apparatus which concerns on embodiment (1) performs. It is the block diagram which showed roughly the principal part of the navigation system by which the speech recognition apparatus which concerns on embodiment (2) was employ | adopted. It is explanatory drawing for demonstrating the installation place of a microphone. It is the flowchart which showed the processing operation which the speech processing part in the speech recognition apparatus which concerns on embodiment (2) performs. It is the block diagram which showed schematically the principal part of the navigation system by which the speech recognition apparatus which concerns on embodiment (3) was employ | adopted. It is explanatory drawing for demonstrating the installation place of a microphone and a CCD camera. It is the flowchart which showed the processing operation which the speech processing part in the speech recognition apparatus which concerns on embodiment (3) performs. It is the flowchart which showed the processing operation which the speech processing part in the speech recognition apparatus which concerns on another embodiment performs.

Explanation of symbols

1, 21, 31 Voice recognition device 2, 22, 23, 32 A / D converter 3, 24, 33 Buffer memory 4 EEPROM
5, 25, 34 Audio processor 6, 26, 27, 36 Microphone 7 Navigation device 35 Image processor 37 CCD camera

Claims (5)

In a speech recognition apparatus that recognizes speech emitted from a speech input target person,
First determination means for determining presence or absence of utterance by a voice input target person;
Second determination means for determining the presence or absence of utterance by a person other than the voice input target person;
A condition establishment determination means for determining whether or not a voice recognition start condition for starting voice recognition is satisfied;
The voice recognition start condition is determined to be uttered by a voice input target person in the first determination means;
The speech recognition apparatus characterized in that the second determination means does not determine that there is a utterance by a person other than the speech input subject until a predetermined period has elapsed since the speech by the speech input subject. . Based on the personality information included in the voice input by the voice input means, the voice judgment means for judging whether or not the voice is the voice input target person,
The speech recognition according to claim 1, wherein the determination by the first determination unit and the determination by the second determination unit are performed based on a determination result by the main utterance determination unit. apparatus. Based on the sound source direction obtained from the voice input by the voice input means, comprises a speaker main judgment means for judging whether or not the voice owner is a voice input target person,
The speech recognition according to claim 1, wherein the determination by the first determination unit and the determination by the second determination unit are performed based on a determination result by the main utterance determination unit. apparatus. Based on the appearance of the voice input target obtained from the image input means, the person other than the voice input target, or the voice input target and the person other than the voice input target, the speaker speaks. It has a utterance main judgment means for judging whether or not it is a target person,
The speech recognition according to claim 1, wherein the determination by the first determination unit and the determination by the second determination unit are performed based on a determination result by the main utterance determination unit. apparatus. When it is determined by the second determination means that a person other than the voice input target person has uttered before the predetermined period has elapsed since the voice input target person uttered,
The speech recognition apparatus according to claim 1, further comprising a holding period setting unit that sets a holding period in which the voice recognition start condition is not satisfied.

Images