JP4005203B2 - In-vehicle speech recognition device - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an in-vehicle voice recognition apparatus that improves noise recognition by reducing noise other than the voice of a speaker.
[0002]
[Prior art]
FIG. 6 is a block diagram showing a system configuration of a conventional in-vehicle speech recognition apparatus. Hereinafter, it demonstrates according to a figure.
In order to reduce the burden on the driver due to the switch operation when operating navigation devices or audio devices while driving a vehicle, etc., it recognizes the voice spoken by the driver (speaker) and inputs instructions to the connected device There is a voice recognition device.
[0003]
Reference numeral 11 denotes an omnidirectional microphone that converts speech uttered by a speaker (speaker) into an electrical signal. Reference numeral 31 denotes a gain adjustment unit for adjusting the signal level detected by the microphone 11 and outputting an appropriate level of audio signal to the audio recognition unit 8. Reference numeral 8 denotes a voice recognition unit connected to an input unit such as a navigation device or an audio device. A control unit 2 controls the gain adjustment unit 31 based on the detection result of the microphone 11.
[0004]
In order to improve the speech recognition rate, the control unit 2 gains the gain adjustment unit 31 based on the level of the speech signal of the speaker detected by the microphone 11 so that a signal having an optimum level is input to the speech recognition unit 8. To control the gain to be adjusted.
[0005]
[Problems to be solved by the invention]
In the in-vehicle voice recognition device, in addition to the voice of the speaker, other passengers' conversations and noises associated with traveling of the vehicle also enter the microphone. Further, the direction of the speaker and the distance between the speaker and the microphone change depending on the physique of the speaker, the posture while driving, and the like. As a result, there is a problem that the noise and signal level input to the voice recognition unit change and the voice recognition rate is lowered.
[0006]
As a countermeasure, there is a method of using a close-talking microphone in which a microphone is fixed to the speaker's mouth, but there is a problem that the speaker, particularly the driver, feels uncomfortable.
An object of the present invention is to provide an in-vehicle speech recognition apparatus that improves noise recognition rate by reducing noise other than the speech of a speaker.
[0007]
[Means for Solving the Problems]
In order to achieve the above object, the present invention provides a direction detecting means for detecting the direction of a speaker, a voice detecting means for detecting a voice of a speaker, and a directivity characteristic of the voice detecting means detected by the direction detecting means. And a voice recognition means for recognizing the voice of the speaker detected by the voice detection means.
[0008]
Further, the voice detection means is composed of a plurality of microphones, and the direction detection means detects the direction of the speaker based on a signal level difference between detection signals detected by the plurality of microphones. It is characterized by being.
Further, the voice detection means is composed of a plurality of microphones, and the direction detection means detects the direction of the speaker based on a delay time between detection signals detected by the plurality of microphones. It is characterized by this.
[0009]
In addition, a position detection unit that detects the position of the speaker is provided, and the direction detection unit determines the direction of the speaker based on the installation position of the voice detection unit and the position of the speaker detected by the position detection unit. It is what is detected.
In addition, a seat setting state detection unit that detects a setting state of the speaker's seat is provided, and the position detection unit is configured to detect the speaker's seat based on the setting state of the speaker's seat detected by the seat setting state detection unit. It is characterized by detecting the position.
[0010]
Further, the voice detecting means is composed of a plurality of microphones, and the directivity adjusting means is a gain adjusting means for performing weighting by adjusting gains of detection signals detected by the plurality of microphones. To do.
Further, the present invention is characterized in that a microphone having a tubular body disposed on the front surface for detecting the voice of the speaker and voice recognition means for recognizing the voice of the speaker detected by the microphone are provided.
The directivity adjusting means selectively enhances the directivity of the direction of the speaker in at least the driver seat or the passenger seat of the vehicle.
[0011]
【Example】
FIG. 1 is a block diagram showing a system configuration of an in-vehicle speech recognition apparatus according to a first embodiment of the present invention. 2A and 2B are diagrams showing speaker position detection and directivity adjustment using a plurality of microphones. FIG. 2A is a diagram showing a positional relationship between a microphone (vertical arrangement) and a speaker. FIG. 2B is a microphone (horizontal arrangement). It is a figure which shows the positional relationship of a speaker. Hereinafter, it demonstrates according to a figure. In this example, the direction of the speaker is detected from the output states of a plurality of microphones, and the gain is adjusted so that the directivity of the voice detection unit 1 is maximized in the direction of the speaker.
[0012]
Reference numeral 1 denotes a voice detection unit (voice detection means) that converts a driver's or passenger's voice into an electrical signal, and includes a plurality of omnidirectional microphones 11, 12, and 13 arranged at predetermined intervals. Reference numeral 31 denotes a gain adjusting unit for individually adjusting the outputs of the microphones 11, 12, and 13 to provide directivity in the direction of the speaker as the entire sound detecting unit 1, and is configured by an electronic volume or the like. Reference numeral 8 denotes a voice recognition unit connected to an input unit such as a navigation device or an audio device. The operation of the navigation device or the audio device is controlled according to the result recognized by the voice recognition unit 8. A control unit 2 detects the direction of the speaker based on the output level or delay time of the microphones 11, 12, and 13 and controls the gain adjustment unit 31 that adjusts the output of each microphone 11, 12, and 13. A memory 21 stores the direction of the speaker detected by the voice detection unit 1 and the adjustment value of the gain adjustment unit 31 in association with each other.
[0013]
First, a method for detecting the direction of the speaker using the microphones 11, 12, and 13 will be described. 2A, signal levels V11, V12, and V13 and delay times t11, t12, and t13 are detected in the three microphones 11, 12, and 13 for the voice uttered by the speaker 81. In this case, the delay time may be a relative delay time based on any one microphone (for example, the center microphone 12) by detecting the rising edge of the signal. When the output characteristics of the microphones are adjusted in advance to be the same, the signal level decreases according to the distance between the speaker 81 and the microphones 11, 12, and 13. The delay time becomes longer in proportion to the difference between the distance between the speaker 81 and the microphones 11, 12, and 13. Therefore, if the positional relationship between the microphones 11, 12, and 13 is known, the direction of the speaker can be calculated.
[0014]
The relationship between the direction of the speaker 81, the signal levels V11, V12, V13 from the microphones 11, 12, 13 and the delay times t11, t12, t13 is stored in the memory 21 in advance. Then, the signal level and delay time are actually detected by the microphones 11, 12, and 13, and the speaker direction is calculated using the relationship between the signal level and delay time stored in the memory 21 and the speaker direction. To do.
[0015]
Next, the directivity adjustment method of the sound detection unit 1 will be described. As shown in FIG. 2A, the three microphones 11, 12, and 13 are arranged in the vertical direction with respect to the speaker, and the gains to be adjusted for the signal levels of the microphones 11, 12, and 13 are G11, G12, and G13. When the gains of the microphones 11, 12, and 13 are adjusted to be the same (G11 = G12 = G13), the directivity of the voice detector 1 is such that the direction of the speaker 81 is strong and the diagonal or lateral direction is weak. it can.
[0016]
Further, as shown in FIG. 2B, the three microphones 11, 12, and 13 are arranged laterally with respect to the speaker, and are adjusted so that the gain increases in the order of the microphones 11, 12, and 13 (G11 <G12 < G13) Then, the directivity of the voice detection unit 1 can be such that the direction of the speaker 81a becomes stronger. On the other hand, by adjusting the gains in the order of the microphones 13, 12, and 11 (G13 <G12 <G11), the directivity of the voice detection unit 1 can be increased in the direction of the speaker 81 b. Furthermore, when the gain of the microphone 12 is adjusted to be higher than the gains of the microphones 11 and 13 (G11 ≦ G12 ≧ G13), the directivity of the voice detection unit 1 is such that the direction of the speaker 81c in the center is stronger. it can.
[0017]
In advance, gains G11, G12, and G13 that should be adjusted so that the directivity of the speaker direction and the speaker direction in the gain adjusting unit 31 are maximized are stored in the memory 21, and the speaker direction can be detected. In the event that the output signal of the microphones 11, 12, 13 is adjusted, the gain adjustment corresponding to the direction of the speaker is performed. In this way, the voice signal in which the voice detection unit 1 in the direction of the speaker is enhanced and the directivity in the other direction is suppressed to reduce the noise from the passenger and the noise from the vehicle is the voice recognition unit. Thus, the recognition rate of the speaker's voice is improved.
[0018]
In addition, a compact unit (corresponding to the sound detection unit 1) in which three microphones 11, 12, and 13 are combined together in one housing may be configured and installed on the back of the map lamp or room lamp in the vehicle. Further, the three microphones 11, 12, and 13 may be installed at locations that are separated from each other (for example, inside the left and right front pillars and the back of the room lamp). If the three microphones are separated from each other, the direction of the speaker can be easily identified.
[0019]
Further, in the microphone arrangement of FIG. 2B, the directivity of the voice detection unit 1 composed of the three microphones 11, 12, and 13 is previously adjusted so as to be maximized in the direction of the driver seat (speaker 81b). The memory 21 stores two types of gains (G11, G12, G13) and gains that should be adjusted so as to be maximized in the direction of the passenger seat (speaker 81a). For any of the speakers (81a, 81b), the direction of the speaker is identified, and the gain is adjusted to the detection signal of the voice detection unit 1 for the speaker, so that the direction of the speaker has directivity. Can be made. That is, a single voice detection unit 1 can perform input processing with a high recognition rate for navigation devices and audio devices from either the driver's seat or the passenger seat.
[0020]
As described above, in this embodiment, the direction of the speaker is detected by a plurality of microphones, and the directivity of the voice detection unit is adjusted to be maximum in the direction of the speaker based on the detection result. The voice signal of the passenger is large, noise such as passengers' conversation and vehicle noise can be reduced, and a voice recognition device with a high degree of recognition can be achieved.
FIG. 3 is a block diagram showing a system configuration of the on-vehicle speech recognition apparatus according to the second embodiment of the present invention. FIG. 4 is a diagram for explaining a seat slide position, a seat surface height, and a backrest angle. Hereinafter, it demonstrates according to a figure. In this example, the direction of the speaker is detected from the setting state of the driver's seat, and the directivity adjustment of the voice detection unit 1 after detecting the direction of the speaker is the same as in the first embodiment. is there.
[0021]
4 is a seat state detection unit (seat setting state detection means) for detecting the setting state of the driver's seat (or passenger seat) of the vehicle, a slide position sensor 41 for detecting the position of the seat in the front-rear direction, and the vertical direction of the seat The seat surface height sensor 42 for detecting the position of the backrest and the backrest angle sensor 43 for detecting the angle of the backrest. The slide position sensor 41 and the seat surface height sensor 42 are sensors for detecting the movement amount (length). For example, a combination of a slit and an optical sensor measures the number of pulses generated according to the movement amount. The backrest angle sensor 43 is a sensor that detects an angle, and measures, for example, a resistance value (or voltage value) that changes according to a rotation angle by using a volume or the like. If the seat setting state can be adjusted electrically (electric seat), the sensor 41, 42, 43 is not provided, and the driving amount of the motor that sets (moves) the seat is measured. The measured drive amount may be substituted. Reference numeral 2 denotes a control unit that detects the position of the driver (speaker) based on the detection result of the seat state detection unit 4 and detects the direction of the speaker from the voice detection unit 1 and the position of the driver. The control unit 2 has a memory 21 in which the detection results of the voice detection unit 1 and the seat state detection unit 4 and the direction of the speaker are stored in advance. A driver's seat 71 includes a seat surface portion 71a, a backrest portion 71b, and a headrest portion 71c. Since the voice detection unit 1, microphones 11, 12, and 13, the control unit 2, the memory 21, the voice recognition unit 8, and the gain adjustment unit 31 have the same names, functions, and operations as the first embodiment, the same numbers are used. The description is omitted.
[0022]
The optimal seat state when the driver is seated in advance, that is, the seat slide position D1, the seat surface height H1, and the backrest angle θ1 are determined by the slide position sensor 41, the seat surface height sensor 42, and the backrest angle sensor 43. The direction of the speaker with respect to the position detection unit 1 installed between the ceiling 70b of the vehicle and the upper portion of the windshield 70a is stored in the memory 21.
[0023]
Next, the state of the seat adjusted when the driver is seated on the seat is detected by the slide position sensor 41, the seat surface height sensor 42, and the backrest angle sensor 43, and the relationship between the seat state and the direction stored in the memory 21 is detected. To detect the direction of the speaker.
When the direction of the speaker is detected, the directivity in the direction of the speaker is maximized by adjusting the gains G11, G12, and G13 corresponding to the microphones 11, 12, and 13, as in the first embodiment. As a result, even if the driver's physique and posture are different and the seat position and the backrest angle change, the optimum microphone directivity can always be obtained.
[0024]
As described above, in this embodiment, the direction of the speaker is detected based on the seat position, and the directivity of the voice detection unit is adjusted to be maximum in the direction of the speaker based on the detection result. Therefore, it is possible to reduce noises such as passengers' conversation and vehicle noise, and a speech recognition device with high recognition can be obtained.
FIG. 5 is a diagram for explaining the installation state of the microphone of the on-vehicle speech recognition apparatus according to the third embodiment of the present invention. Hereinafter, it demonstrates according to a figure. In this example, a non-directional microphone is used so that the directivity of the voice detection unit is maximized in the direction of the speaker.
[0025]
By installing the microphone 11 on the interior material surface 72a of the vehicle via the pipe 75, the sound in the direction limited by the opening diameter φ and the length L of the pipe 75 is selectively detected by the microphone 11, and the other directions Can be prevented from being detected. By adjusting the direction of the pipe 75 in advance so as to coincide with the direction of the speaker, the conversation and noise of the passenger can be reduced.
[0026]
As described above, in this embodiment, the voice detector is installed so that the directivity of the voice detection unit is maximized in the direction of the speaker, so that the speaker's voice signal is large and noise such as passenger conversation and vehicle noise is reduced. And a speech recognition device with a high degree of recognition.
[0027]
【The invention's effect】
As described above, the present invention can provide a vehicle-mounted speech recognition apparatus that improves noise recognition rate by reducing noise other than the speaker's speech.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a system configuration of an in-vehicle speech recognition apparatus according to a first embodiment of the present invention.
FIG. 2 is a diagram illustrating speaker position detection and directivity adjustment using a plurality of microphones.
FIG. 3 is a block diagram showing a system configuration of an on-vehicle speech recognition apparatus according to a second embodiment of the present invention.
FIG. 4 is a diagram showing a seat surface height, a slide position, and a backrest angle of a seat.
FIG. 5 is a diagram showing a microphone installation state of the on-vehicle speech recognition apparatus according to the third embodiment of the present invention.
FIG. 6 is a block diagram showing a system configuration of a conventional in-vehicle speech recognition device.
[Explanation of symbols]
1... Voice detector 43... Back angle sensor,
11, 12, 13... Microphone, 2... Control unit,
4 ... Seat state detection unit, 21 ... Memory,
41... Slide position sensor 31.
42... Seat height sensor, 8.

Claims (6)

Direction detecting means for detecting the direction of the speaker;
A plurality of voice detection means for detecting the voice of the speaker;
A directional characteristic adjusting unit that adjusts a directional characteristic of the voice detecting unit by adjusting a gain of detection signals detected by the plurality of voice detecting units to perform weighted synthesis ;
Storage means for storing the direction of the speaker in correspondence with the gain adjustment value for increasing the directivity of the voice detection means in the direction of the speaker;
Voice recognition means for recognizing the voice of the speaker detected by the voice detection means ,
The in-vehicle speech recognition apparatus characterized in that the directivity adjusting means adjusts the directivity of the voice detecting means based on a gain adjustment value corresponding to the direction of the speaker stored in the storage means . The voice detection means is composed of a plurality of microphones,
The in-vehicle voice recognition according to claim 1, wherein the direction detecting means detects the direction of the speaker based on a signal level difference between detection signals detected by the plurality of microphones. apparatus. The voice detection means is composed of a plurality of microphones,
2. The on-vehicle speech recognition apparatus according to claim 1, wherein the direction detecting means detects the direction of the speaker based on a delay time between detection signals detected by the plurality of microphones. . Comprising position detecting means for detecting the position of the speaker;
The said direction detection means detects the direction of the said speaker based on the installation position of the said audio | voice detection means, and the position of the speaker detected by the said position detection means. In-car speech recognition device. A seat setting state detecting means for detecting a setting state of a speaker's seat;
5. The in-vehicle voice according to claim 4, wherein the position detecting means detects the position of the speaker based on the seat setting state of the speaker detected by the seat setting state detecting means. Recognition device. 2. The on-vehicle speech recognition apparatus according to claim 1, wherein in the storage means, the direction of the speaker is at least a driver seat and a passenger seat.

Images