JP3919337B2 - Voice recognition device for vehicle - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a vehicle voice recognition apparatus, and more particularly to detection of a driver's voice.
[0002]
[Prior art]
In recent years, with the spread of car navigation, a vehicle voice recognition device has been used to set a destination. Further, in a hands-free telephone or the like, a vehicular voice recognition device is used for setting a telephone number.
[0003]
In general, in a running vehicle interior, there are noises such as conversations in the front passenger seat and rear seats, sound from car audio, or noise from outside the vehicle when a window is opened. Is currently not good.
[0004]
In view of this, improvements have been made to devices that remove these noises and detect only the voice of the operator.
In a conventional vehicle voice recognition device, the steering pad is provided with a microphone for recognizing the voice of the driver. This microphone has a fairly sharp directivity for detecting the voice of the driver only, and the direction of the directivity is directed to the pilot's lips.
[0005]
[Problems to be solved by the invention]
However, since the directivity is sharp, the direction of the directivity easily deviates from the position of the lips of the pilot depending on the attitude of the pilot, and stable voice is not input at present.
[0006]
So, in the noise in the passenger compartment Also, It is required that the voice of the operator can be stably input to the voice recognition device.
It is an object of the present invention to be installed in a noisy vehicle interior. And An object of the present invention is to provide a vehicle voice recognition device capable of stably inputting a voice of a driver.
[0007]
[Means for Solving the Problems]
In order to solve the above problem, the invention according to claim 1 is a voice recognition means for recognizing a driver's voice based on a plurality of microphones provided in the vehicle and a voice signal detected by the plurality of microphones. In the vehicle speech recognition apparatus comprising: the plurality of microphones. Is , Steering wheel Le As well as Steering delay time is defined as the maximum delay time of the synthesized speech signal composed of multiple microphones, and the steering delay time relative to the turning position of the steering wheel is stored at the preset lip position of the driver. Storage means for Based on the output signal of the rotational position detecting means for detecting the rotational position of the steering wheel, the rotational position of the steering wheel is calculated, With reference to the storage contents of the storage means At the turning position of the steering wheel at that time Opposite Steering delay time The Delay time calculation to set Stepped Delayed voice for synthesizing voice signals of the plurality of microphones by delaying at least one voice signal of the plurality of microphones from a voice signal of another microphone by a steering delay time set by the delay time calculation means Synthetic hand Stepped The main point is that
[0015]
According to the invention of claim 1, The delay time calculation means calculates the rotation position of the steering wheel based on the output signal of the rotation position detection means for detecting the rotation position of the steering wheel, and at the rotation position of the steering wheel at that time, a plurality of At least one audio signal from the microphone In the pilot's lip position set in advance A delay time for maximizing a synthesized voice signal that is delayed and synthesized from the voice signal of the microphone and the voice signal of another microphone is set as the steering delay time.
[0016]
And The delayed speech synthesizer is configured to add at least one speech signal of the plurality of microphones to the steering delay time. The ma Since the voice signals of the plurality of microphones are synthesized after being delayed from the voice signal of the microphone, the synthesized voice signal can best extract the voice from the direction of the lip position of the pilot.
[0017]
Therefore, even if the steering wheel is rotated, the voice of the driver is always stably supplied to the voice recognition means. As a result, since the speech recognition means performs speech recognition based on the synthesized speech signal, the speech recognition rate can be improved.
[0021]
DETAILED DESCRIPTION OF THE INVENTION
(First embodiment)
DESCRIPTION OF EMBODIMENTS Hereinafter, an embodiment of a vehicle voice recognition device embodying the present invention will be described with reference to FIGS.
[0022]
As shown in FIG. 1, a pillar (A pillar) 2 in front of a driver's seat in a passenger compartment 1 is provided with a CCD camera 3 as an imaging means so as to focus on a driver's face.
[0023]
FIG. 2 shows the steering wheel 4 in a state where it is not rotated in the left-right direction as viewed from the driver seat (neutral state). First and second microphones 5 and 6 are provided apart from each other at the peripheral edge of the steering wheel 4. In this embodiment, the microphones 5 and 6 are installed on the steering wheel 4 so that the distance between them is 34 centimeters.
FIG. 3 shows an electrical configuration of the vehicle speech recognition apparatus configured as described above.
[0024]
The CCD camera 3 outputs an image signal (video signal) including the lips of the driver's face. This image signal is input to the image processing device 11. The image processing apparatus 11 includes an image digital signal processor (image DSP) 12 as a lip position calculation means, a read / write memory (image RAM) 13 for temporarily storing data, and a read only memory (image ROM). 14 and a high-pass filter 15. The image ROM 14 stores a control program for performing high-speed digital computation by the image DSP 12. This control program is transferred to the image RAM 13 in the image DSP 12 and a desired image recognition process is performed in the image DSP 12.
[0025]
The high-pass filter 15 differentiates the input image signal and outputs a differentiated image signal. The image DSP 12 receives the differential image signal from the high-pass filter 15 and calculates the contour of the driver's face. The image DSP 12 calculates the center position of a rectangle circumscribing the outline of the face as the nose position. The image DSP 12 calculates a lip position by adding a preset distance to the nose position. The image DSP 12 outputs a lip position signal P1 relative to the lip position to the sound processing device 16.
[0026]
The first and second microphones 5 and 6 output the voice of the operator to the voice processing device 16. The voice processing device 16 includes a voice DSP 17, a voice RAM 18, and a voice ROM 19 as delay time calculation means, delayed voice synthesis means, and voice recognition means. The audio ROM 19 stores a control program for performing high-speed digital computation by the audio DSP 17. This control program is transferred to the audio RAM 17 in the audio DSP 16 and desired audio recognition processing is performed in the audio DSP 16. The voice ROM 19 stores a plurality of standard voice command patterns composed of a plurality of preset voice command patterns. Furthermore, the voice ROM 19 stores data of the delay time τ for each lip position (lip position signal P1) of the operator when sitting in the driver's seat.
[0027]
The delay time τ is a time for delaying the audio signal of the second microphone 6 from the audio signal of the first microphone 5. The audio signal of the second microphone 6 is delayed by this delay time τ in the audio DSP 17. The audio signal of the first microphone 5 is synthesized in the audio DSP 17 with the audio signal of the second microphone 6 delayed by the delay time τ to become a synthesized audio signal. That is, the directivity when the first microphone 5 and the second microphone 6 are considered as one microphone is changed by delaying and synthesizing the audio signal of one of the first microphone 5 and the second microphone 6. It is known. Therefore, the directivity of the microphone that can collect the voice of the pilot with the highest sensitivity is obtained by delaying the voice signal of the second microphone 6 with a delay time τ that matches the lip position of the pilot. Can be set.
[0028]
FIG. 4 (a) shows the period when the frequency of the audio signal is 1000 Hz as T (= 0.001 second), its wavelength as λ, and the distance between the first and second microphones 5 and 6 as λ / 2 (= The directivity characteristics of the synthesized speech signal at a delay time τ (= T / 2 = 0.0005 seconds) in the case of 34 centimeters) are shown. Assuming that the delay time τ is T / 2, as shown in FIG. 4A, the synthesized speech signal in the delay time τ (= T / 2 = 0.0005 seconds) is from 0 degrees and 180 degrees. The voice of the pilot is extracted best. In other words, the delay time τ of T / 2 is such that when the operator's lip position is in front of the steering wheel 4 in the state shown in FIG. It is time to obtain the most excellent directivity for.
[0029]
Similarly, FIG. 4B shows the directivity characteristics of the synthesized speech signal at the delay time τ (= T / 3≈0.0003 seconds). Assuming that the delay time τ is T / 3, as shown in FIG. 4B, the synthesized speech signal at the delay time τ (= T / 3≈0.0003 seconds) is from 340 degrees and 200 degrees. The voice of the pilot is extracted best. That is, the delay time τ of T / 3 is the first and second microphones when the operator's lip position is tilted 20 degrees from the front to the left with respect to the steering wheel 4 in the state shown in FIG. Times 5 and 6 are times for obtaining the best directivity with respect to the voice of the pilot.
[0030]
Similarly, FIG. 4C shows the directivity characteristics of the synthesized speech signal at the delay time τ (= 3T / 5 = 0.006 seconds). Assuming that the delay time τ is 3T / 5, as shown in FIG. 4C, the synthesized speech signal in the delay time τ (= 3T / 5 = 0.006 seconds) is from 10 degrees and 170 degrees. The voice of the pilot is extracted best. That is, the delay time τ of 3T / 5 is the first and second microphones when the operator's lip position is tilted 10 degrees from the front to the right with respect to the steering wheel 4 in the state shown in FIG. Times 5 and 6 are times for obtaining the best directivity with respect to the voice of the pilot.
[0031]
Therefore, the delay time τ of each lip position with respect to each of the microphones 5 and 6 stored in the voice ROM 19 is the most excellent directivity characteristic when the synthesized voice signal is emitted by the operator at the current lip position. It is the time that was made available.
[0032]
The voice DSP 17 reads the plurality of standard voice command patterns and the delay time τ data into the voice RAM 18 in the voice DSP 17 together with the start of the vehicle by the ignition switch.
[0033]
The audio DSP 17 receives the lip position signal P1 from the image DSP 12, and reads out the delay time τ for the lip position from the audio RAM 17 based on the lip position signal P1.
[0034]
The audio DSP 17 inputs audio signals from the first and second microphones 5 and 6, and the audio DSP 17 delays the audio signal of the second microphone 6 based on the delay time τ.
[0035]
The audio DSP 17 synthesizes the audio signals from the first and second microphones 5 and 6. The voice DSP 17 starts voice recognition based on the synthesized voice signals of both microphones 5 and 6, and calculates a voice pattern (actual voice pattern). When the actual voice pattern matches any one of a plurality of preset standard voice command patterns, the voice DSP 17 outputs a control signal relative to the matched standard voice command pattern.
[0036]
Now, for example, when the lip position of the driver is at a 0 degree position, the image DSP 12 outputs a lip position signal P1 relative to the lip position (= 0 degree). The audio DSP 17 reads the delay time τ (= T / 2) for the current lip position from the audio RAM 17 based on the lip position signal P1. The audio DSP 17 synthesizes the audio signals of the first and second microphones 5 and 6 by delaying the audio signal of the second microphone 6 by the delay time τ (= T / 2) from the audio signal of the first microphone 5. To do. Therefore, as shown in FIG. 4A, the synthesized voice signal is most extracted from the direction of the lips position (= 0 degree) of the driver.
[0037]
According to the vehicle voice recognition apparatus of the above embodiment, the following features can be obtained.
(1) In the present embodiment, the audio DSP 17 calculates a delay time τ for delaying the audio signal of the second microphone 6 based on the lip position of the operator with respect to the microphones 5 and 6. The audio DSP 17 synthesizes the audio signals of the first and second microphones 5 and 6 by delaying the audio signal of the second microphone 6 by the delay time τ from the audio signal of the first microphone 5. In the synthesized voice signal synthesized in this way, the voice from the direction from the lips position of the driver is extracted most.
[0038]
Therefore, the voice of the pilot is always stably supplied to the voice DSP 17 even if the position of the lip of the pilot changes. As a result, since the speech DSP 17 performs speech recognition based on the synthesized speech signal, the speech recognition rate can be improved.
[0039]
(2) In the present embodiment, the face of the driver who sits in the driver's seat and emits sound is imaged by the CCD camera 3, and the lip position of the driver is recognized by the image SDP 12. Therefore, it is possible to accurately detect which direction the driver is always speaking.
[0040]
(Second Embodiment)
A second embodiment in which the present invention is embodied in a vehicle voice recognition device will be described below with reference to FIG.
[0041]
The configuration of this embodiment is different from that of the first embodiment in that a steering angle sensor 20 is provided in place of the CCD camera 3 and the image processing device 11 of the vehicle voice recognition device of the first embodiment.
[0042]
A steering shaft (not shown) is provided with a steering angle sensor 20 (not shown) as a turning position detecting means for detecting the turning position of the steering wheel 4. The steering angle sensor 20 is composed of an optical rotary encoder.
[0043]
FIG. 5 shows the electrical configuration of the vehicle speech recognition apparatus.
The first and second microphones 5 and 6 output the voice of the operator to the voice processing device 16.
[0044]
The voice processing device 16 includes a voice DSP 17, a voice RAM 18, and a voice ROM 19 as delay time calculation means, delayed voice synthesis means, and voice recognition means. The audio ROM 19 stores a control program for performing high-speed digital computation by the audio DSP 17. This control program is transferred to the audio RAM 18 in the audio DSP 17 and desired audio recognition processing is performed in the audio DSP 17. The voice ROM 19 stores a plurality of standard voice command patterns composed of a plurality of preset voice command patterns. Further, the voice ROM 19 stores data of a plurality of steering delay times τs set in advance in association with a plurality of rotation positions of the steering wheel 4 in one lip position of the pilot set in advance.
[0045]
The steering delay time τs is a time for delaying the audio signal of the second microphone 6. The plurality of steering delay times τs are delay times for maximizing a synthesized voice signal obtained by synthesizing the voice signals of the first and second microphones 5 and 6 at the turning position of the steering wheel 4 at that time.
[0046]
As the vehicle is started by the ignition switch, the plurality of standard voice command patterns and the data of the delay time τ are read into the voice RAM 18 in the voice DSP 17.
[0047]
The steering angle sensor 20 outputs a signal relative to the rotational position of the steering wheel 4.
The audio DSP 17 inputs a signal from the steering angle sensor 20, calculates the rotational position of the steering wheel 4, and reads the steering delay time τs relative to the rotational position from the audio RAM 18. The audio DSP 17 inputs the audio signals of both microphones 5 and 6 at the rotation position, and the audio signal of the second microphone 6 is input to the audio signal of the first microphone 5 in the audio DSP 17 using the delay time τs. The synthesized voice signal is generated by synthesizing the voice signals of both microphones 5 and 6 with a further delay.
[0048]
The voice DSP 17 starts voice recognition based on the synthesized voice signals of the microphones 5 and 6, calculates a real voice pattern, and the real voice pattern is one of a plurality of standard voice command patterns set in advance. When it matches, the control signal corresponding to the matched standard voice command pattern is output.
[0049]
Now, the voice DSP 17 inputs a signal from the steering angle sensor 20 to calculate the rotational position of the steering wheel 4 at one predetermined lip position of the operator. The audio DSP 17 reads the steering delay time τs relative to the rotation position from the audio RAM 18. Then, the audio DSP 17 receives the audio signals of the two microphones 5 and 6 at the rotation position, and delays the audio signal of the second microphone 6 from the audio signal of the first microphone 5 by the delay time τs. A synthesized speech signal is generated by synthesizing 5 and 6 speech signals.
[0050]
Therefore, the synthesized voice signal best extracts the voice from the direction of the lips position of the driver at the turning position.
Next, when the steering wheel 4 is rotated, the audio DSP 17 inputs a signal from the steering angle sensor 20 and calculates the rotation position of the steering wheel 4 at that time. The sound DSP 17 reads the steering delay time τs relative to the rotation position from the sound RAM 18. Then, the audio DSP 17 receives the audio signals of the two microphones 5 and 6 at the rotation position, and delays the audio signal of the second microphone 6 from the audio signal of the first microphone 5 by the delay time τs. A synthesized speech signal is generated by synthesizing 5 and 6 speech signals.
[0051]
Therefore, the synthesized voice signal best extracts the voice from the direction of the operator's lip position even at the turning position during the steering operation. That is, the voice DSP 17 can best extract the voice from the direction of the operator's lips even when the steering wheel 4 rotates.
[0052]
According to this embodiment, the following features can be obtained.
(1) The voice DSP 17 inputs a signal from the steering angle sensor 20 and calculates the rotational position of the steering wheel 4. The audio DSP 17 reads the steering delay time τs relative to the rotation position from the audio RAM 18. Then, the audio DSP 17 receives the audio signals of the two microphones 5 and 6 at the rotation position, and delays the audio signal of the second microphone 6 from the audio signal of the first microphone 5 by the delay time τs. A synthesized speech signal is generated by synthesizing 5 and 6 speech signals.
[0053]
The synthesized voice signal generated in this way best extracts the voice from the direction of the operator's lip position at the occasional turning position of the steering wheel 4 at one preset lip position of the operator. ing.
[0054]
Therefore, the audio DSP 17 The Even if the tearing wheel 4 rotates, the voice of the driver is always supplied stably. As a result, the voice DSP 17 can improve the voice recognition rate.
[0055]
(2) In this embodiment, since the CCD camera 3 and the image DSP 12 as in the first embodiment are unnecessary, the cost can be reduced.
(Third embodiment)
A third embodiment in which the present invention is embodied in a vehicle voice recognition device will be described below with reference to FIG.
[0056]
The configuration of the present embodiment is obtained by removing the steering angle sensor 20 from the vehicle voice recognition device of the second embodiment.
The first and second microphones 5 and 6 output the voice of the operator to the voice processing device 16.
[0057]
The voice processing device 16 includes a voice DSP 17, a voice RAM 18, and a voice ROM 19 as delay time calculation means, delayed voice synthesis means, and voice recognition means. The audio ROM 19 stores a control program for performing high-speed digital computation by the audio DSP 17. This control program is transferred to the audio RAM 18 in the audio DSP 17 and desired audio recognition processing is performed in the audio DSP 17. The voice ROM 19 stores a plurality of standard voice command patterns composed of a plurality of preset voice command patterns.
[0058]
When the voice signal of the first microphone 5 that outputs the voice of the driver and the voice signal of the second microphone 6 delayed by the delay time τ are synthesized, the synthesized voice signals of the first and second microphones have directivity and the directivity thereof. It is known that the synthesized voice signal also changes because the characteristics change depending on the value of the delay time τ. Therefore, by calculating the delay time τ that maximizes the synthesized voice signal and setting the delay time τ to the delay time τ of the second microphone 6, it is possible to best extract the voice from the driver's lip position. it can.
[0059]
As the vehicle is started by the ignition switch, the plurality of standard voice command patterns are read into the voice RAM 18 in the voice DSP 17.
The audio DSP 17 inputs the audio signals of the first and second microphones 5 and 6 and sequentially changes the delay time τ within a preset range, and the audio signal of the second microphone 6 is changed in various ways in the audio DSP 17. Delayed from the audio signal of the first microphone 5 by the delay time τ. Then, the audio DSP 17 synthesizes the audio signals of both microphones 5 and 6 at each delay time τ to generate a synthesized audio signal. The speech DSP 17 compares these synthesized speech signals, and calculates a delay time τ (= τm) when the synthesized speech signal is maximized.
[0060]
The audio DSP 17 sets the delay time τm to the calculated delay time τm, delays the audio signal of the second microphone 6 by the delay time τm from the audio signal of the first microphone 5, and the audio signals of both microphones 5 and 6. Are synthesized to generate a synthesized speech signal.
[0061]
The voice DSP 17 starts voice recognition based on the synthesized voice signals of the microphones 5 and 6, calculates a real voice pattern, and the real voice pattern is one of a plurality of standard voice command patterns set in advance. When it matches, the control signal corresponding to the matched standard voice command pattern is output.
[0062]
According to this embodiment, the following features can be obtained.
(1) In the present embodiment, the audio DSP 17 inputs audio signals from the first and second microphones 5 and 6 and sequentially changes the delay time τ within a preset range, so that the audio signal of the second microphone 6 is obtained. Is delayed from the audio signal of the first microphone 5 by a delay time τ, and the audio signals of the first and second microphones 5 and 6 are synthesized to generate a synthesized audio signal. The audio DSP 17 calculates a delay time τm that maximizes the synthesized audio signal. That is, the delay time τm is set to a value that best extracts the voice of the driver.
[0063]
The voice DSP 17 synthesizes the voice signals of the two microphones 5 and 6 by delaying the voice signal of the second microphone 6 by the delay time τm from the voice signal of the first microphone 5. Regardless of the lip position of the operator, the sound from the direction of the lip position can be extracted best.
[0064]
Accordingly, the directivity of the microphones 5 and 6 is always directed to the direction of the lip position of the pilot, so that the voice of the pilot is always supplied stably. As a result, since the speech DSP 17 performs speech recognition based on the synthesized speech signal, the speech recognition rate can be improved.
[0065]
(2) In this embodiment, since the CCD camera 3 and the image DSP 12 as in the third embodiment and the steering angle sensor 20 as in the second embodiment are unnecessary, the cost can be reduced.
[0066]
In addition, you may change embodiment of this invention as follows.
In the first embodiment, an imaging tube may be used instead of the CCD camera 3. In the first embodiment, a plurality of CCD cameras 3 may be used instead of one CCD camera 3. In this case, the image signal of one CCD camera 3 that accurately captures the operator's lips among the plurality of CCD cameras 3 is selected in the image DSP 12, and the image DSP 12 is based on the selected image signal. To calculate the pilot's lip position.
[0067]
In the first embodiment, the CCD camera 3 is provided in the A pillar 2, but the CCD camera 3 may be provided anywhere in the passenger compartment 1 where an image of the driver's face can be obtained.
In the first embodiment, the period T is set to 0.001 seconds, but any numerical value may be used as long as it is a frequency period in the audio frequency range (several hertz to 20 kilohertz).
[0068]
In the second embodiment, an optical rotary encoder is used to detect the rotational position of the steering wheel 4, but this may be changed to a magnetic rotary encoder.
[0069]
In the second embodiment, the voice ROM 19 is associated with a plurality of rotational positions of the steering wheel 4 at a preset lip position of the operator, and a delay time τ composed of a plurality of preset steering delay times τs. However, data of delay time τ composed of a plurality of preset steering delay times τs may be stored in association with a plurality of rotational positions of the steering wheel 4 at a plurality of lip positions of the operator. .
[0070]
In the first to third embodiments, one or more microphones may be further provided on the steering pad 4 in addition to the first and second microphones 5 and 6.
In this configuration, the audio DSP 17 calculates each delay time τ for delaying the added microphone audio signal, delays the audio signal of each microphone by the respective delay time τ, and then sets each microphone. The audio signal is synthesized.
[0071]
In each of the above embodiments, the first microphone 5 and the second microphone 6 are used. However, three, four, or more may be used.
In each of the above embodiments, the first and second microphones 5 and 6 are installed on the steering wheel 4. However, the first and second microphones 5 and 6 may be installed at other locations other than the steering wheel 4, such as an instrument panel.
[0072]
In each of the above embodiments, at least one of the image DSP 12 and the audio DSP 17 may be replaced with a central processing unit (CPU).
Even when configured as in each of the above examples, substantially the same features can be obtained in the above embodiments.
[0073]
Next, technical ideas other than the invention described in the claims that can be grasped from the respective embodiments and other examples will be described below together with their effects.
(1) In a voice detection method for a vehicle voice recognition device comprising a plurality of microphones provided in a vehicle and voice recognition means for recognizing a voice of a driver detected by the plurality of microphones.
A delay time (τ) for delaying at least one audio signal of the plurality of microphones is calculated based on the lip position of the driver, and at least one of the audio signals of the plurality of microphones is delayed. A speech detection method for a vehicle speech recognition apparatus that detects a driver's speech by synthesizing speech signals of a plurality of microphones, delayed by time (τ) from a speech signal of another microphone.
[0074]
In this case, the voice of the driver is always stably supplied to the voice recognition means. As a result, since the speech recognition means performs speech recognition based on the synthesized speech signal, the speech recognition rate can be improved.
[0075]
【The invention's effect】
As described above in detail, according to the first aspect of the present invention, even if the steering wheel is rotated, Is Longitudinal voice is always supplied stably. As a result, since the speech recognition means performs speech recognition based on the synthesized speech signal, the speech recognition rate can be improved.
[Brief description of the drawings]
FIG. 1 is a schematic view of a passenger compartment for explaining the arrangement of a CCD camera and a microphone.
FIG. 2 is a schematic diagram of a steering pad for explaining the arrangement of microphones.
FIG. 3 is an electrical configuration diagram of the vehicle voice recognition device according to the first embodiment.
FIGS. 4A and 4B are directivity characteristics diagrams of a synthesized voice signal of both microphones, where FIG. 4A is a delay time (= T / 2), FIG. 4B is a delay time (= T / 3), and FIG. ) Shows the case of delay time (= 3T / 5).
FIG. 5 is an electrical configuration diagram of a vehicle voice recognition device according to a second embodiment.
FIG. 6 is an electrical configuration diagram of a vehicle voice recognition device according to a third embodiment.
[Explanation of symbols]
τ, τm ... delay time, τs ... steering delay time, P1 ... lip position signal, 3 ... imaging means, 4 ... steering wheel, 5 ... 1st microphone as a microphone, 6 ... 2nd microphone as a microphone, 12 ... Image DSP as lip position calculating means, 17 ... Speech recognition means, A voice DSP as a delay voice synthesis means, a delay time calculation means, 20... A steering angle sensor as a rotation position detection means.

Claims (1)

A vehicle equipped with a plurality of microphones (5, 6) provided in the vehicle and voice recognition means (17) for recognizing the voice of the driver based on the voice signals detected by the plurality of microphones (5, 6). In a speech recognition device,
The plurality of microphones (5, 6) are provided on the steering wheel (4),
The delay time that maximizes the synthesized voice signal obtained by synthesizing the voice signals of the plurality of microphones (5, 6) is defined as the steering delay time (τs). Storage means (19) for storing the steering delay time (τs) relative to the rotational position;
Based on the output signal of the rotation position detecting means (20) for detecting the rotation position of the steering wheel (4), the rotation position of the steering wheel (4) is calculated, and the storage contents of the storage means (19) are stored. referring to the delay time calculating means for setting the relative steering delay (.tau.s) the rotational position of the steering wheel (4) at that time (17),
At least one audio signal of the plurality of microphones (5, 6) is delayed from the audio signals of other microphones by a steering delay time (τs) set by the delay time calculating means (17). A speech recognition apparatus for a vehicle, comprising delay speech synthesis means (17) for synthesizing speech signals of the microphones (5, 6).

Images