JP2007065122A - Noise suppressing device of on-vehicle voice recognition device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To effectively suppress noise so that a voice recognition rate can be improved in various operation states of each device of a vehicle. <P>SOLUTION: The noise suppressing device for suppressing a voice signal which is input to a voice recognition device in order to recognize voice of a speaker in the vehicle, comprises: a microphone 1 provided in the vehicle in order to collect internal sound of the vehicle; a vehicle state detection section 3 for detecting a vehicle state which is an operation state of each device of the vehicle; a memory section 11 for storing the sound signal collected with the microphone 1 by relating it to the vehicle state when collected; a vehicle sound to vehicle state comparison memory section 13 for storing information in which the stored sound signal for each vehicle state, is averaged in a predetermined period; and a vehicle sound suppressing section 14 for suppressing the sound by subtracting the averaged information corresponding to the vehicle state when the sound signal is collected, from the sound signal collected with the microphone 1. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a noise suppression device that suppresses the sound of a vehicle from the voice of a speaker in the vehicle.

In recent vehicles, an in-vehicle voice recognition device that controls equipment in the vehicle by voice generated by an occupant may be used. However, since the sound generated by the vehicle itself or the sound generated from the audio device provided in the vehicle becomes noise and is superimposed on the sound generated by the occupant, the sound may not be recognized correctly. Therefore, a method for suppressing noise has been proposed as preprocessing for speech recognition.
Patent Document 1 discloses a music signal from a car audio device mixed in a sound input microphone from an audio signal output from the car audio device and a detection signal detected from a sound input microphone provided in the vicinity of a speaker. A technique for estimating and removing this from the detection signal is described. The technique of Patent Document 1 is a noise reduction method as preprocessing for voice recognition. However, since the driving sound of an automobile is not reduced, there is a high possibility that it can be handled only when the driving sound is low, such as when the vehicle is stopped.
Patent Document 2 describes a method of suppressing noise so as to match the state of noise generated according to the running state of the vehicle. In the technique of Patent Document 2, the subtraction control unit of the noise suppression device controls the opening and closing of the switch based on the vehicle information obtained from the vehicle information detection device, so that the voice signal data of the speaker input to the voice microphone The subtractor removes stationary noise and non-stationary noise from the subtractor according to the traveling state of the vehicle. Noise suppression is performed in accordance with four patterns of the vehicle running state: stopping, acceleration / deceleration, engine stop, and running at a constant speed. However, the traveling state includes not only these four patterns but also various states, and the traveling sound of the vehicle changes according to the traveling state. Therefore, noise suppression with a fixed pattern as shown in Patent Document 2 may not be sufficient.
Patent Document 3 describes a method of providing a cut-off frequency determination unit and a high-pass filter controlled by the cut-off frequency determination unit after the voice detection unit. The cut-off frequency determination unit analyzes the main subject of noise in each band of the audio signal. If the main noise is periodic noise, the band is set as a cut-off frequency in the high-pass filter. Thereby, even if the band of periodic noise changes, it is always reduced. When random noise is the main component, a predetermined cutoff frequency is set. As a result, the human voice band is not cut to reduce the recognition rate.
However, when the cut-off frequency of the high-pass filter is the upper limit, a part of the voice information is also affected depending on the speaker. Under certain noise, the recognition rate may be increased even if some of the voice information is lost. However, in the sentence, “Vowels important for human speech recognition are voiced sounds, and the voiced sounds are 100 to 300 Hz. It is a periodic sound when it has a fundamental frequency in the band ”, whereas in the description of the embodiment, there is a contradiction in that the cutoff frequency is raised to 400 Hz. The technique of Patent Document 3 is not effective against noise superimposed on a frequency band necessary for speech recognition.
JP 2000-231399 A JP 2000-321080 A JP 2001-296877 A

  The present invention has been made in view of the above circumstances, and an object thereof is to effectively suppress noise so as to improve a speech recognition rate in various operating states of each device of a vehicle.

  In order to achieve the above object, a noise suppression apparatus according to a first aspect of the present invention is a noise suppression apparatus that suppresses noise of a voice signal input to a voice recognition apparatus in order to recognize a voice of a speaker in a vehicle. An apparatus for collecting acoustic signals in the vehicle for collecting sound inside the vehicle, vehicle state detecting means for detecting a vehicle state as an operating state of each device of the vehicle, and A storage unit that stores the acoustic signal collected by the acoustic signal collecting unit in association with the vehicle state at the time of collection, and information obtained by averaging the acoustic signal stored for each vehicle state over a predetermined period Corresponding to the vehicle state when the acoustic signal is collected from the vehicle sound-vehicle state comparison storage unit that stores the vehicle signal in association with the vehicle state and the acoustic signal collected by the acoustic signal collecting unit. The averaged Characterized in that it comprises a suppressing means for suppressing subtracted distribution, the.

  Further, only acoustic signals having a phase difference equal to a difference in distance between the plurality of acoustic signal collecting means, the position of a specific speaker to be recognized by the speech recognition apparatus, and the plurality of acoustic signal collecting means. Specific speaker phase signal extraction means for extracting from the acoustic signal, and the specific speaker phase signal extraction means, from the acoustic signals collected by the plurality of acoustic signal collection means, By extracting a component having a phase difference equal to the difference in distance from each of the plurality of acoustic signal collection means, the component has a phase different from the difference in distance between the specific speaker and each of the plurality of acoustic signal collection means. The acoustic component is suppressed.

  In the case where the vehicle is provided with an audio device, it is preferable to include an audio signal suppression unit that subtracts the sound signal of the audio device from the sound signal collected by the sound signal collecting device to obtain an sound signal outside the vehicle. .

  Preferably, the acoustic signal collected by the acoustic signal collecting means is digitized and processed as frequency information and / or time axis information.

  As the operating state of each device of the vehicle, vehicle speed, accelerator opening, engine speed, acceleration / deceleration state, transmission state, steering steering angle, wiper operating state, air conditioner operating state, electric fan operating state, window opening / closing state Any combination of the horn operating state and the type of tire mounted on the vehicle may be included.

  According to the noise suppression device of the on-vehicle speech recognition device of the present invention, an acoustic information database (vehicle sound-vehicle state comparison memory) corresponding to various operating states of each device of the vehicle is constructed and collected by the acoustic signal collecting means. Since the component of the vehicle running sound that is characteristic of the host vehicle corresponding to the state of the vehicle when the acoustic signal is collected is suppressed from the generated acoustic signal, only the voice of the speaker in the vehicle can be captured. As a result, an improvement in the speech recognition rate can be expected in the in-vehicle speech recognition apparatus.

(Embodiment 1)
An embodiment of a noise suppression device 30 according to the present invention will be described with reference to FIG. FIG. 1 is a block diagram showing a noise suppression device 30 according to Embodiment 1 of the present invention. The configuration of each part of the noise suppression device 30 shown in FIG. 1 will be described. The microphone 1 is an acoustic signal collecting unit that collects sound in the vehicle. The acoustic signal collected by the microphone 1 is amplified by the microphone amplifier 4, converted into a digital signal by the A / D converter 5, and input to the audio signal suppression unit 9. The audio signal input unit 2 inputs an audio signal of an audio device (not shown) provided in the vehicle. The audio signal is converted into a digital signal by the A / D converter 6 and input to the audio signal adjusting unit 8. The sampling frequency in the A / D converters 5 and 6 is several to 10 times the maximum frequency effective for speech recognition. For example, when the maximum frequency for recognizing sound is 10 kHz, the sampling frequency is about 40 kHz to 50 kHz.

  The vehicle state detection unit 3 detects the operation state of each device of the vehicle. The operating state of each device of the vehicle includes, for example, vehicle speed, accelerator opening, engine speed, acceleration / deceleration state, transmission state, steering steering angle, wiper operating state, air conditioner operating state, electric fan operating state, window There are open / closed states, horn operating states, and types of tires mounted on the vehicle. In order to detect the operating state of each device of the vehicle, operating state information is input from a vehicle control device (not shown). Alternatively, each device described above is provided with a detector to detect its state. Alternatively, operating state information is input from the control unit of each device of the vehicle. The vehicle state information detected by the vehicle state detection unit 3 is transmitted to the audio signal adjustment unit 8 and the address generation unit 10.

  The audio signal adjustment unit 8 is composed of, for example, a variable band filter that can change the amplification degree of the audio signal for each band, and adjusts the audio signal to be subtracted and suppressed from the acoustic signal outside the vehicle according to the vehicle state. For example, the acoustic reflection conditions change depending on the open / close state of the window, and the magnitude of the sound of the audio device in the car that wraps around the microphone 1 changes, so the amplification level of the audio signal to be subtracted and suppressed is adjusted according to the open / close state of the window To do. The audio signal adjusted by the audio signal adjusting unit 8 is input to the audio signal suppressing unit 9.

  The audio signal suppression unit 9 includes a subtraction circuit for obtaining a difference between acoustic signals and a Fourier transform circuit, and subtracts and suppresses the audio signal adjusted by the audio signal adjustment unit 8 from the acoustic signal collected by the microphone 1. For the suppression of the audio signal, the difference can be obtained with the time axis information (sound waveform). After suppressing the audio signal from the acoustic signal with the time axis information, the frequency information is obtained. First, the acoustic signal and the audio signal may be Fourier-transformed to obtain frequency information, and the difference may be obtained from the frequency information. In order to obtain frequency information, an acoustic signal and an audio signal are accumulated for a certain period of time, and Fourier transform is performed at each certain time interval. Thereafter, processing is performed at regular intervals.

  The acoustic signal in which the audio signal is suppressed is sent to and stored in the storage unit 11. The acoustic signal stored in the storage unit 11 is averaged over a predetermined period for each vehicle state by the averaging processing unit 12. The sound signal is averaged for each frequency. The vehicle sound-vehicle state comparison storage unit 13 stores the vehicle sound obtained by averaging the acoustic signals over a predetermined period for each vehicle state by the averaging processing unit 12.

  The averaging processing unit 12 weights and averages the averaged vehicle sound corresponding to the vehicle state at that time and the current acoustic signal, and the vehicle sound − as the averaged vehicle sound corresponding to the vehicle state The storage in the vehicle state comparison storage unit 13 is updated. For example, assuming that N is a positive integer, the averaged sound is multiplied by N-1 and the sound signal is added and divided by N to obtain a new averaged vehicle sound. For example, when N is 100, the acoustic signal first becomes 1/100, and thereafter becomes 99/100 each time, and the contribution rate to the averaged vehicle sound decreases gradually.

  As an averaging method, a moving average of a predetermined number of acoustic signals having the same vehicle state may be taken. For averaging, the difference between the current acoustic signal (with the audio signal suppressed) and the averaged vehicle sound may be subtracted from the current acoustic signal, but it may be averaged even in the same vehicle condition in the long term. Since the sound may change, the audio signal may be averaged with the suppressed acoustic signal. In this way, for example, when a tire is replaced, it is possible to follow seasonal variations or long-term changes.

  The vehicle sound-vehicle state comparison storage unit 13 is a non-volatile memory such as a flash memory, a hard disk, a DVD (Digital Versatile Disc), a DVD-RAM (Digital Versatile Disc Random-Access Memory), a DVD-RW (Digital Versatile Disc Rewritable). Consists of

  FIG. 2 is a diagram illustrating an example of the structure of the vehicle sound-vehicle state comparison storage unit 13. A bit is assigned to each item of the vehicle state, and an address where the averaged vehicle sound is stored corresponding to the bit pattern determined by the state of 0 or 1 of each bit (vehicle sound storage location) Is assigned. Each bit is determined by the vehicle state information, and an averaged vehicle sound corresponding to the vehicle state can be extracted with reference to the vehicle sound storage location of the row that matches the bit pattern.

  The vehicle state includes vehicle speed (vehicle speed), transmission stage, accelerator opening, engine speed, acceleration / deceleration, steering angle, wiper operation, air conditioner operation, electric fan operation, window opening, There is a horn operating state. For example, the vehicle speed is a numerical value that increases by 1 every 10 km / h. Since the state of the transmission is 8 stages including, for example, 5 forward stages, 1 reverse stage, neutral and parking, 3 bits are allocated. Similarly, for example, 3 bits are assigned to the accelerator opening, 4 bits are assigned to the engine speed, and 3 bits are assigned to the acceleration / deceleration state. In the wiper operation, for example, 2 bits are assigned as 4 stages of stop, intermittent, continuous, and high speed. In the air conditioner operating state, for example, a total of 3 bits are allocated to the operation of the fan of the heat pump and the heat exchanger and the operation of the indoor fan. For example, 2 bits are assigned to the operating state of the electric fan. As the opening of the window, for example, 3 bits are assigned in 8 stages including the fully closed state as a whole. For example, 2 bits may be allocated for each window. If the horn is only ON / OFF, 1 bit is sufficient. In the example of FIG. 2, the vehicle state is 31 bits in total. The items constituting the vehicle state and the number of bits can be added or changed as appropriate according to the vehicle. For example, raindrop sensor information may be added to add the strength of rain to one of the vehicle conditions.

  The address generation unit 10 generates an address in the vehicle sound-vehicle state comparison storage unit 13 in which the averaged vehicle sound corresponding to the vehicle state is stored, and instructs the vehicle sound-vehicle state comparison storage unit 13. For example, 31-bit information corresponding to the vehicle state of FIG. 2 is generated from the vehicle state information detected by the vehicle state detection unit 3, and the vehicle sound storage location corresponding to the generated 31-bit information is shown in FIG. Extract from the data represented by the structure.

  The vehicle sound stored in the vehicle sound storage location instructed by the address generation unit 10 is input from the vehicle sound-vehicle state comparison storage unit 13 to the vehicle sound suppression unit 14. In addition, an acoustic signal stored in the storage unit 11 is input to the vehicle sound suppression unit 14. The vehicle sound suppression unit 14 includes a subtraction circuit, and subtracts the vehicle sound averaged from the acoustic signal for each frequency component. That is, in the first embodiment, the vehicle sound suppression unit 14 suppresses the averaged vehicle sound corresponding to the vehicle state at that time from the acoustic signal.

  FIG. 3 is a schematic diagram illustrating an example of suppressing vehicle sound averaged from an acoustic signal. FIG. 3A shows a state in which sound is superimposed on the vehicle sound. For convenience, the lower solid line indicating the vehicle sound and the upper solid line indicating the superimposed sound are separated, but the observed data is the upper solid line. In the example of FIG. 3, the case where an acoustic signal is Fourier-transformed into frequency information is shown. (B) of FIG. 3 shows the acoustic component other than the own vehicle, which is the remainder obtained by subtracting (suppressing) the spectrum of the vehicle sound from the spectrum of (a). In this manner, the vehicle sound component can be suppressed from the acoustic signal, and the sound component in the vehicle can be extracted. Strictly speaking, the suppressed acoustic signal includes sound outside the vehicle, for example, the sound of a vehicle traveling nearby, which is reduced by the sound insulation characteristics of the vehicle.

  The suppression of the vehicle sound averaged from the acoustic signal is performed at regular intervals. The fixed time interval is set to a short interval that allows voice recognition. That is, processing is performed at a short time interval that can detect a change between phonemes and phonemes. For example, processing is performed at time intervals of 10 milliseconds or less, preferably 5 milliseconds or less.

  The acoustic signal in which the vehicle sound averaged by the vehicle sound suppression unit 14 is suppressed is input to the speech recognition device 15 and recognized. When the time axis information is required, the signal waveform of the frequency component frequency may be synthesized from the frequency information of the acoustic signal whose vehicle sound is suppressed by the vehicle sound suppression unit 14. In that case, the phase of the composite waveform is determined using the phase information of the original acoustic signal.

  Of the noise suppression device 30, the address generation unit 10, the audio signal suppression unit 9, the audio signal adjustment unit 8, the storage unit 11, the averaging processing unit 12, and the vehicle sound suppression unit 14 are all or part of a DSP (Digital (Signal Processor).

  Next, the operation of the noise suppression device 30 of FIG. 1 will be described with reference to FIG. FIG. 4 is a flowchart showing the operation of the noise suppression apparatus 30 according to the first embodiment.

  First, acoustic signals in the vehicle are collected by the microphone 1, and an audio signal is input by the audio signal input unit 2 (step A1). At the same time, the vehicle state at that time is input by the vehicle state detector 3 (step A2). As described above, the audio signal and the audio signal are digitized by the A / D converters 5 and 6, the audio signal adjusting unit 8 adjusts the audio signal according to the vehicle state, and the audio signal suppressing unit 9 converts the audio signal from the audio signal to the audio signal. The signal is suppressed (step A3). The acoustic signal in which the audio signal is suppressed is stored in the storage unit 11. From the vehicle state information, an address of the vehicle sound-vehicle state comparison storage unit 13 in which the averaged vehicle sound corresponding to the vehicle state is stored is generated, and the address corresponding to the vehicle state is generated from the vehicle sound-vehicle state comparison storage unit 13. Reference is made to the averaged vehicle sound (step A4).

  The vehicle sound suppression unit 14 suppresses the averaged vehicle sound corresponding to the vehicle state from the acoustic information in which the audio signal is suppressed (step A5), and the sound recognition device 15 receives the acoustic signal that suppresses the vehicle sound. Send (step A6).

  The averaging processing unit 12 weights and averages the averaged vehicle sound corresponding to the vehicle state at that time and the current vehicle sound, and the vehicle sound − as the averaged vehicle sound corresponding to the vehicle state The storage of the vehicle state comparison storage unit 13 is updated (step A7). The processing so far is performed at predetermined time intervals (for example, 5 milliseconds). The averaged vehicle sound other than the vehicle state at that time is not updated at that time.

  As a result, an acoustic information database (vehicle sound-vehicle state comparison memory) corresponding to various operating states of each device of the vehicle was constructed, and acoustic signals were collected from the acoustic signals collected by the acoustic signal collecting means. Since the characteristic vehicle running sound component of the host vehicle corresponding to the state of the vehicle at the time is suppressed, only the voice of the speaker in the vehicle can be captured. As a result, an improvement in the speech recognition rate can be expected in the in-vehicle speech recognition apparatus.

(Embodiment 2)
A second embodiment of the present invention will be described with reference to FIGS. In the second embodiment, the noise suppression device includes a plurality of acoustic signal collection units, and the position of the specific speaker to be recognized by the speech recognition device from the acoustic signals collected by the plurality of acoustic signal collection units, and the plurality of the plurality of acoustic signal collection units. Only acoustic signals having a phase difference equal to the difference in distance from each of the acoustic signal collecting means are extracted. FIG. 5 is a plan view of a vehicle showing an example of the arrangement of acoustic signal collecting means according to Embodiment 2 of the present invention. As shown in FIG. 5, for example, microphones 1 a and 1 b can be provided as acoustic signal collecting means at two locations on the front of the driver.

  FIG. 6 is a block diagram of the noise suppression apparatus 30 according to the second embodiment. Compared with the first embodiment, there are a plurality (two) of microphones 1a and 1b. The acoustic signal is amplified by the microphone amplifiers 4a and 4b for each of the microphones 1a and 1b, and converted into a digital signal by the A / D converters 5a and 5b. A specific speaker phase signal extraction unit 7 is newly provided.

  The specific speaker phase signal extraction unit 7 is configured by a DSP, for example. The specific speaker phase signal extraction unit 7 extracts a signal component having a phase difference corresponding to the difference between the distance from the microphone 1a to the speaker and the distance from the microphone 1b to the speaker. For example, the signal of the microphone 1b is shifted by the time corresponding to the arrival time of the sound wave, which is the difference between the distance between the microphone 1a and the speaker, and the distance between the microphone 1b and the speaker, and the correlation between the two signals is obtained. Then, a highly correlated part is extracted. At this time, the amplitude is corrected according to the ratio of the distance between the microphones 1a and 1b and the speaker. When the distances between the microphones 1a and 1b and the speakers to be recognized by voice, for example, the driver, are equal, the arrival times of the voices from the speakers to the microphones 1a and 1b are equal, so there is no need to shift the time of the acoustic signal. .

  A signal having a phase difference corresponding to the difference between the distance from the microphone 1a to the speaker and the distance from the microphone 1b to the speaker, extracted by the specific speaker phase signal extraction unit 7, is again used as an audio signal as an audio signal. Input to the suppression unit 9. After the audio signal suppression, the audio signal is suppressed by the audio signal suppression unit 9 and the vehicle sound averaged by the vehicle sound suppression unit 14 is suppressed and sent to the voice recognition device 15 as in the first embodiment. Further, the averaging processing unit 12 weights and averages the averaged vehicle sound corresponding to the vehicle state at that time and the current acoustic signal, and the vehicle sound is obtained as the averaged vehicle sound corresponding to the vehicle state. -Update the storage of the vehicle state comparison storage unit 13. In the case of the second embodiment, the averaged vehicle sound becomes a component having the same phase difference as the difference in distance from the microphones 1a, 1b to the speaker among the acoustic signals collected by the microphones 1a, 1b. ing.

  Next, the operation of the noise suppression device 30 of FIG. 6 will be described with reference to FIG. FIG. 7 is a flowchart for explaining the operation of the noise suppression apparatus 30 according to the second embodiment. 7, step B1 and step B2 are similar to steps A1 and A2 of FIG. 2, in-vehicle acoustic signals are collected by the microphones 1a and 1b, and audio signals are input by the audio signal input unit 2 (step B1). . At the same time, the vehicle state at that time is input by the vehicle state detector 3 (step B2). In the second embodiment, a specific speaker phase signal is extracted before audio signal suppression (step B3). Steps B4 to B8 after audio signal suppression are the same as steps A3 to A7 in FIG.

  In the present embodiment, a plurality of microphones (acoustic signal collecting means) 1a and 1b are provided, and a signal having a phase difference corresponding to the difference between the distance from the microphone 1a to the speaker and the distance from the microphone 1b to the speaker. Since the components are extracted, signals having other phase differences are suppressed, and unnecessary speaker's voice and vehicle sound are suppressed. Further, since signals with different phase differences are suppressed from the sound outside the vehicle, there is a further effect of noise suppression as compared with the first embodiment.

  Next, a modification of the second embodiment will be described. FIG. 8 is a block diagram illustrating a different example of the second embodiment. In the noise suppression device 30 of FIG. 8, the storage unit 11, the averaging processing unit 12, and the vehicle sound suppression unit 14 of FIG.

  The microcomputer 20 includes a control unit 21, a main storage unit 22, an external storage input / output unit 23 and an input / output unit 24. The main storage unit 22, the external storage input / output unit 23, and the input / output unit 24 are all connected to the control unit 21 via the internal bus 25. In the configuration of FIG. 8, the vehicle sound-vehicle state comparison storage unit 13 is an external storage unit of the microcomputer 20.

  The control unit 21 of the microcomputer is composed of a CPU (Central Processing Unit) or the like, and receives an acoustic signal according to a program stored in an external storage unit, and executes vehicle sound suppression and averaging processing.

  The main storage unit 22 includes a RAM (Random-Access Memory) and the like, and is used as a work area for the control unit 21.

  The external storage unit including the vehicle sound-vehicle information comparison storage means 13 includes a flash memory, a hard disk, a DVD (Digital Versatile Disc), a DVD-RAM (Digital Versatile Disc Random-Access Memory), and a DVD-RW (Digital Versatile Disc Rewritable). A program for causing the control unit 21 to perform the above processing is stored in advance, and in accordance with instructions from the control unit 21, the program and other data used by the program are stored in the control unit 21. The data supplied from the control unit 21 is stored. The external storage unit includes a vehicle sound-vehicle state comparison storage unit 13, and supplies an averaged vehicle sound corresponding to the vehicle state and updates a newly averaged vehicle sound according to an instruction from the control unit 21. And remember.

  The input / output unit 24 includes a serial interface, a parallel interface, or a LAN (Local Area Network) interface. The control unit 21 inputs an acoustic signal from the audio signal suppression unit 9 via the input / output unit 24 and outputs data to be transmitted to the voice recognition device 15.

  In the second embodiment, the specific speaker phase signal extraction unit 7, the audio signal adjustment unit 8, the audio signal suppression unit 9, the address generation unit 10, and the vehicle sound suppression unit 14 of the noise suppression device 30 are all or one of them. The unit can be configured by a DSP (Digital Signal Processor). That is, the DSP may include a specific speaker phase signal extraction unit 7, an audio signal adjustment unit 8, an audio signal suppression unit 9, an address generation unit 10, and a microcomputer 20.

  Since each process is described by a program by using a microcomputer or a DSP, parameters such as time interval adjustment and phase difference can be changed, and it becomes easy to deal with different vehicles.

It is a block diagram of the noise suppression apparatus which concerns on Embodiment 1 of this invention. It is a figure which shows the example of the structure of a vehicle sound-vehicle state comparison memory | storage part. It is a schematic diagram which shows the example of a mode that a vehicle sound is suppressed from an acoustic signal. 3 is a flowchart showing the operation of the first embodiment. 6 is a plan view showing an example of arrangement of acoustic signal collecting means according to Embodiment 2. FIG. 5 is a block diagram of a noise suppression device according to Embodiment 2. FIG. 10 is a flowchart showing the operation of the second embodiment. FIG. 10 is a block diagram illustrating a different example of the second embodiment.

Explanation of symbols

1, 1a, 1b microphone (acoustic signal collecting means)
2 Audio signal input section
3 Vehicle state detector 5, 5a, 5b A / D converter
6 A / D converter
7 Specific speaker phase signal extraction unit
9 Audio signal suppressor
11 Memory unit
12 Averaging processor
13 Vehicle sound-vehicle state comparison storage unit
14 Vehicle sound suppression part
15 Voice recognition device
30 Noise suppressor

Claims (5)

A noise suppression device that suppresses noise in a voice signal input to a voice recognition device in order to recognize a speaker's voice in a vehicle,
Acoustic signal collecting means provided in the vehicle for collecting the sound inside the vehicle;
Vehicle state detecting means for detecting a vehicle state that is an operating state of each device of the vehicle;
A storage unit that stores the acoustic signals collected by the acoustic signal collecting unit in association with the vehicle state when collected;
A vehicle sound-vehicle state comparison storage unit that stores information obtained by averaging the acoustic signals stored for each vehicle state over a predetermined period in association with the vehicle state;
Suppression means for subtracting and suppressing the averaged information corresponding to the vehicle state when the acoustic signal is collected from the acoustic signal collected by the acoustic signal collecting means;
A noise suppression device comprising: A plurality of acoustic signal collecting means;
A specific speaker that extracts only an acoustic signal having a phase difference equal to a difference in distance between a position of a specific speaker to be recognized by the speech recognition apparatus and each of the plurality of acoustic signal collecting means from the acoustic signal. Phase signal extraction means;
With
The specific speaker phase signal extraction unit has a phase difference equal to a difference in distance between the specific speaker and each of the plurality of acoustic signal collection units from the acoustic signal collected by the plurality of acoustic signal collection units. The noise suppression according to claim 1, wherein an acoustic component having a phase different from a difference in distance between the specific speaker and each of the plurality of acoustic signal collecting units is suppressed by extracting the component. apparatus.   When the vehicle includes an audio device, the vehicle further includes an audio signal suppression unit that suppresses the sound signal of the audio device from the sound signal collected by the sound signal collection device to obtain the sound signal inside the vehicle. Item 2. The noise suppression device according to Item 1.   The noise suppression device according to claim 1, wherein the acoustic information of the noise suppression device is digitized and processed as frequency information and / or time axis information.   As the operating state of each device of the vehicle, vehicle speed, accelerator opening, engine speed, acceleration / deceleration state, transmission state, steering steering angle, wiper operating state, air conditioner operating state, electric fan operating state, window opening / closing state The noise suppression device according to claim 1, comprising any combination of a horn operating state and a type of tire mounted on the vehicle.

Images