JP2000321080A - Noise suppressor, voice recognizer and car navigation system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To remove noise component appropriately from a voice input signal even under an environment where noise generation state varies significantly. SOLUTION: The subtraction control section 44 of a noise suppressor 32 controls subtraction and removal of steady noise α1 and unsteady noise r(t) from the voice signal data SN(t) of a speaker inputted to a voice microphone through subtractors 35, 41 depending on the traveling state of an automobile by controlling the switching of switches 37, 40, 43 based on vehicle information obtained from a vehicle information detector. Voice recognition rate of a voice recognizer is enhanced by suppressing noise to conform to the generating state depending on the traveling state thereby suppressing strain being imparted to a voice data s(t) as much as possible.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a noise reducing apparatus for removing a noise component from an input signal in which a speech signal component to be recognized and a noise signal component are mixed, thereby improving a speech signal recognition rate. The present invention relates to a suppression device, a voice recognition device, and a vehicle navigation device including the voice recognition device.

[0002]

2. Description of the Related Art A speech recognition apparatus is used, for example, in a car navigation apparatus, and recognizes a voice uttered by an occupant of a car to set a destination, for example, and performs setting. In this case, the speech recognition device compares the input speech with a plurality of comparison target patterns stored in advance, and determines the one having the highest degree of coincidence as the recognition result.

[0003] In a car under the actual use environment of a car navigation device, it is difficult to avoid noise components when a voice uttered by a speaker is input. To improve the noise, how to suppress and remove the noise component is an important issue.

Therefore, for example, as one of the methods for removing a noise component from an input signal after determining a voice input section, a spectrum subtraction method (for example, STEVEN F BOL) is used.
L, Suppression of Acoustic Noise in Speech Using Sp
ectral Subtraction, IEEE Tran.on Acoustics, Speech a
nd Signal processing, Vol.Assp-27, No.2 April 1979, p
p.113-120). The spectrum subtraction method divides the input audio data into short-time frames, and subtracts the noise spectrum estimated in advance from the spectrum of the input audio on which noise is superimposed for each frame, to obtain the original audio data. The spectrum is estimated.

[0005]

However, since the noise estimated in advance is estimated by referring to the input state immediately before the input of the voice, the voice to be recognized is actually input. There is a problem that the noise does not coincide with the noise generated in the section. Therefore, this method is effective in suppressing stationary noise that occurs relatively continuously and whose properties do not change much in a short time, such as driving sounds of air conditioners, engines, and motors. Unsteady noise that suddenly occurs and changes its properties in a short time, such as noise from a moving object, cannot be removed.

Therefore, as disclosed in Japanese Patent Application Laid-Open No. 4-245300, a microphone for voice input and a microphone for noise input are separately provided, and a noise component is determined based on data obtained from both. There is one that removes stationary noise and non-stationary noise by estimating and subtracting in real time.

However, for example, in a cabin of an automobile, the state of generation of noise around the automobile changes greatly when the automobile is running and when it is stopped. If noise removal is uniformly performed in such a situation, there is a possibility that excessive distortion may occur in audio data to be properly obtained, and in such a case, on the contrary, the speech recognition rate is reduced. Will be.

The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a noise control apparatus capable of appropriately removing a noise component from a speech input signal even in an environment in which the state of noise generation changes greatly. An object of the present invention is to provide a suppression device, a voice recognition device, and a vehicle navigation device including the voice recognition device.

[0009]

According to the first aspect of the present invention, the first stationary noise subtracting means subtracts the stationary noise estimated based on the input signal of the voice microphone from the input signal, The second stationary noise subtracting means subtracts the stationary noise estimated based on the input signal of the noise microphone from the input signal. Then, in the signal output from the first stationary noise subtracting means, a non-stationary noise component remains in addition to the audio signal. The non-stationary noise estimating means estimates the non-stationary noise based on the output signals of the first and second stationary noise subtracting means, and the control means determines the non-stationary noise performed by the non-stationary noise subtracting means according to the surrounding state. Controls the execution of noise subtraction.

In other words, when extracting the audio signal data, if the nature of the noise generated in the surroundings is mainly stationary, non-stationary noise subtraction does not give distortion to the audio signal data. In some cases, the extraction can be performed well without the need. Therefore, the control means controls the execution of the subtraction of the non-stationary noise according to the surrounding state, so that the noise can be suppressed so as to be adapted to the state.

According to the second aspect of the present invention, the first stationary noise subtracting means subtracts the stationary noise estimated based on the input signal of the voice microphone from the input signal, and subtracts the second stationary noise. The means subtracts stationary noise estimated based on the input signal of the noise microphone from the input signal.
The non-stationary noise estimating unit estimates non-stationary noise based on the input signals of the voice microphone and the noise microphone, and the non-stationary noise subtracting unit extracts the non-stationary noise from the output signal of the first stationary noise subtracting unit. Subtract. The control means controls the execution of the subtraction performed by the first and second stationary noise subtracting means and the non-stationary noise subtracting means according to the surrounding state.

That is, depending on the state of the noise generated in the surroundings, it is preferable to subtract only the stationary noise, to subtract only the non-stationary noise, or not to subtract both of them. In some cases, it can be extracted well. Therefore, the control means controls the execution of the subtraction of the stationary noise and the non-stationary noise according to the surrounding state, so that the noise can be suppressed so as to be more suitable for the state.

According to the third aspect of the present invention, the non-stationary noise estimating means estimates the non-stationary noise based on the input signal of the voice microphone and the input signal of the noise microphone. The stationary noise is estimated based on the input signal of the voice microphone. Then, the non-stationary noise subtracting means subtracts the non-stationary noise from the input signal of the voice microphone, the stationary noise subtracting means subtracts the stationary noise from the output signal of the non-stationary noise subtracting means, and the control means The execution of the subtraction by the non-stationary noise subtracting means and the stationary noise subtracting means is controlled in accordance with the state. Therefore, the same effect as the second aspect can be obtained. Also, when estimating non-stationary noise, the configuration is simplified by subtracting the stationary noise from the input signal beforehand and performing the subtraction of the non-stationary noise in a lump after removing the non-stationary noise. While being able to
The load on the arithmetic processing can be reduced.

According to the fourth or fifth aspect of the present invention, the voice recognizing means analyzes the voice data output from the noise suppressing apparatus and recognizes the voice data pattern. Therefore, it is possible to accurately recognize the data pattern based on the audio data in which noise has been well suppressed.

According to the vehicle navigation apparatus of the sixth aspect, the predetermined processing is executed based on the voice data pattern recognized by the voice recognition means in the voice recognition apparatus of the fifth aspect, so that the noise environment changes. It is possible to recognize the voice of the occupant of the vehicle as a speaker with high accuracy even in a cabin where the traffic is severe. Therefore, the occupant can reliably perform a desired operation by voice input without manually operating an input key or the like, thereby improving convenience and ensuring driving safety.

According to the vehicle navigation device of the present invention, when the vehicle state detecting means detects the state of the vehicle, it outputs a detection signal to the voice recognition device. Then, the control means of the noise suppression device included in the speech recognition device controls the execution of the subtraction by the stationary noise subtracting unit and the non-stationary noise subtracting unit according to the given detection signal. Therefore, the control unit can appropriately control the execution of the subtraction of the stationary noise and the non-stationary noise based on the running state of the vehicle, the operation state of each unit, and the like.

According to the vehicle navigation apparatus of the present invention, the control means of the noise suppression device performs the subtraction only in the non-stationary noise subtraction means when the wiper of the vehicle is operating. Control. That is, in the case of rainy weather, the wiper of the vehicle is normally operated, but in such a case, the intermittent operation sound of rain or the wiper is generated as noise having an unsteady property. Is considered to be dominant. Therefore, by executing only the subtraction in the non-stationary noise subtracting means, the non-stationary noise can be effectively suppressed, and the speech recognition rate can be improved.

According to the vehicle navigation device of the ninth aspect, the control means of the noise suppression device controls so that only the subtraction by the non-stationary noise subtraction means is executed when the vehicle is stopped. . That is, for example, when the vehicle is stopped at a traffic light at an intersection, a traveling sound of a passing vehicle, an operation sound of a turn signal, and the like are generated as noise having an unsteady property. Therefore, by executing only the subtraction in the non-stationary noise subtracting means, the non-stationary noise can be effectively suppressed as in the ninth aspect.

According to a tenth aspect of the present invention, when the vehicle is accelerating or decelerating, only the subtraction by the non-stationary noise subtracting unit is performed when the vehicle is accelerating or decelerating. To control. That is,
When the vehicle accelerates or decelerates, the engine speed changes rapidly, and the engine sound is generated as non-stationary noise. Therefore, by executing only the subtraction in the non-stationary noise subtracting means, it is possible to effectively suppress the non-stationary noise as in the eighth or ninth aspect.

According to the vehicle navigation apparatus of the eleventh aspect, the control means of the noise suppression apparatus includes a stationary noise subtracting means and a non-stationary noise subtracting means in accordance with the vehicle navigation information obtained via the vehicle state detecting means. Since the execution of the subtraction by the subtraction means is controlled, it is possible to determine how to execute the subtraction of the stationary noise and the nonstationary noise in consideration of the surrounding environment where the vehicle is located.

According to the vehicle navigation apparatus of the twelfth aspect, the control means of the noise suppression apparatus includes a stationary noise subtraction means when the vehicle is located in the parking lot and the engine of the vehicle is stopped. And the non-stationary noise subtraction means is controlled not to perform any subtraction. That is, when the vehicle is parked in a parking lot, no engine noise as stationary noise is generated, and traffic of the vehicle around the vehicle is relatively small, so that non-stationary noise arriving from outside is hardly assumed. Therefore, in such a case, it is possible to obtain audio data with less distortion by not subtracting both the stationary noise and the non-stationary noise.

According to the vehicle navigation apparatus of the thirteenth aspect, the control means of the noise suppression apparatus comprises a stationary noise subtraction means and an unsteady noise control means in accordance with traffic information obtained from the VICS receiver via the vehicle state detection means. The execution of the subtraction in the noise subtraction means is controlled. That is, since the frequency of occurrence of the non-stationary noise changes depending on the traffic volume, it is possible to appropriately determine whether to execute the non-stationary noise subtraction by referring to the VICS traffic information.

According to the fourteenth aspect of the present invention, the control means of the noise suppression device determines that the traffic on the road on which the vehicle is traveling at a substantially constant speed and the vehicle is currently traveling. If the difference is equal to or less than the predetermined amount, control is performed so that only the subtraction in the stationary noise subtracting means is performed.

That is, when the vehicle is traveling at a substantially constant speed on a road with a small traffic volume, engine noise, road noise,
The wind noise and the like are generated in a substantially constant state, while the passing noise of other vehicles is hardly generated. Therefore, the former noise having a stationary characteristic is dominant. Therefore, by performing only the subtraction in the stationary noise subtracting means, noise can be effectively suppressed without giving any distortion to the voice data, and the voice recognition rate can be improved.

[0025]

DESCRIPTION OF THE PREFERRED EMBODIMENTS (First Embodiment) A first embodiment in which the present invention is applied to a car navigation system will be described below with reference to FIGS. In FIG. 1 showing the electric configuration of the entire car navigation device 1, the car navigation device 1 includes a position detector 2, a map data input device 3, an operation switch group 4, a control circuit 5 to which these are connected, and the like. . Further, an external memory 6, a display device 7 such as a display, a remote control sensor 8, a voice recognition device 9, and the like are connected to the control circuit 5. The control circuit 5 is configured as a microcomputer,
Although not specifically shown, a well-known CP
U, ROM, RAM, I / O, etc. These components are electrically connected to each other via a bus line.

The position detector 2 is a well-known geomagnetic sensor 1
0, gyroscope 11, distance sensor 12, and GP
A GPS receiver 13 for detecting the position of the vehicle based on radio waves from an S (Global Positioning System) satellite is provided. Each of the sensors 10 to 13 generates an error having a different property.
By referring to the information from, each information is used while making corrections as necessary. Note that, depending on the required accuracy, a configuration may be made using a part of the sensors 10 to 13 or a steering sensor, a wheel sensor, or the like may be added.

The map data input device 3 is a device for inputting various data including map data and landmark data, so-called map matching data for improving position detection accuracy, and the like. CD-R as data storage medium
Although OM is generally used, DVD (Digital Ver.
satile Disk) or a memory card may be used.

On the screen of the display device 7, map data input from the map data input device 3 is displayed, and a marker indicating the current position of the vehicle input from the position detector 2 is displayed on the map data. In addition, additional data such as a guidance route and a mark of a set point are displayed in a superimposed manner.

The car navigation system (car navigation) 1 is provided from a remote control sensor 8 via a remote control terminal (hereinafter referred to as a remote control) 14 or
When the position of the destination is input via the operation switch group 4, a so-called route guidance function for automatically selecting an optimal traveling route from the current position to the destination and displaying the route as a guidance route is provided. I have. As a method for automatically setting an optimum route, for example, the Dijkstra method is known. As the operation switch group 4, for example, a touch switch or a mechanical switch formed integrally with the display device 7 is used, and is used for various input operations.

The voice recognition device 9 is used when the operation switch group 4 and the remote controller 14 are manually operated to indicate a destination or the like, and similarly when the user performs a voice input. It is arranged so that a destination or the like can be instructed.

The voice recognition device 9 includes a voice recognition unit (voice recognition means) 15, a dialogue control unit 16, a voice synthesis unit 17, a voice extraction device 18, a voice microphone 19, and a noise microphone 2.
0, PTT (Push To Talk) switch 21, speaker 22
And a control unit 23.

When the speech recognition unit 15 performs speech recognition processing on the speech data supplied from the speech extraction device 18 according to the instruction of the dialog control unit 16, the speech recognition unit 15 outputs the recognition result.
6 is output. That is, the voice recognition unit 15
Compares and compares the voice data obtained from the voice extraction device 18 with a plurality of comparison target pattern candidates registered in the dictionary data stored in advance, and determines a higher comparison target pattern having a high degree of coincidence with the dialog control unit. 16 is output.

The recognition of a word sequence in the input voice is performed by sequentially analyzing the voice data by acoustic analysis, for example, by LPC (Linear Prediction).
Coefficient) Acoustic features such as cepstrum are extracted, and their time series data is obtained. Then, the obtained time series data is divided into several sections, and each section is stored as dictionary data using a known DP matching method, HMM (Hidden Markov Model), or a method such as a neural network. Ask if they are compatible with

The dialog control unit 16 issues a response voice utterance instruction to the voice synthesizing unit 17 and a control circuit 5 for controlling the entire car navigation system 1 from the voice recognition result and the internal state managed by itself.
For example, a destination required for the navigation process is notified and an instruction to execute the setting process is issued. That is, if this voice recognition device 9 is used,
The user can give a destination instruction or the like to the car navigation system 1 by voice input without operating the operation switch group 4 or the remote controller 14.

The voice microphone 19 is provided for inputting a voice uttered by the user. The noise microphone 20 is provided for inputting noise existing as a background of a voice uttered by the user.

The voice extracting device 18 includes a PTT switch 21
When the control unit 23 detects that the user has been turned on by the user, the control unit 23 converts the voice signal uttered by the driver based on the signals input to the voice microphone 19 and the noise microphone 20 into digital data, as described later. And outputs it to the speech recognition unit 15. The control unit 2
3 gives an instruction to the amplifier 24 of the on-vehicle audio equipment to temporarily reduce the volume of the audio signal output to the audio speaker 25 to substantially zero (mute) or to output the audio signal of the audio signal. It is configured so that an instruction to return the volume to the original state (cancel mute) can be given.

Here, a detailed configuration of the voice extracting device 18 will be described with reference to FIG. Voice extraction device 18
Is composed of a frame division unit 26, a speech section determination unit 27, a speech buffer 28, a noise buffer 29, a Fourier transform unit 30 for speech, a Fourier transform unit 31 for noise, and a noise suppression device 32. .

The frame dividing section 26 includes microphones 19 and 20
The input audio signal is sampled at a rate of, for example, about several tens of KHz and A / D converted, and the obtained discrete data sequence is cut out for each section (frame) set to a time width of about several tens of ms. The signal is divided and multiplied by a window function (for example, a Hanning window) for suppressing a harmonic component generated by an end of the cutout, and is output to the voice section determination unit 27.

The voice section judging section 27 includes the voice microphone 19
It is determined whether or not each frame of the audio signal input to the audio buffer includes a voice signal uttered by the speaker. The data of the frame including the audio signal is determined by the audio buffer 28. To be stored. However, in order to estimate a steady-state noise component to be described later, a section containing only noise is accumulated at the head of the audio buffer 28 for one frame.

Further, the voice section determination unit 27 determines, among the respective frame data of the acoustic signal input to the noise microphone 20,
The data of the frame corresponding to the frame containing the audio signal (that is, at the same time) is output to the noise buffer 29 and accumulated.

In the following, digital data of the 19 audio microphones after the frame division section 26 will be referred to as audio signal data, and digital data of the 20 noise microphones will be referred to as miscellaneous audio signal data.

The Fourier transform units 30 and 31 include a buffer 2
The data stored in each of the sections 8 and 29 is subjected to a fast Fourier transform (FFT) to generate a short-time power spectrum and output to the noise suppression device 32. The noise suppression device 32 includes a short-time spectrum (spectrum including voice and noise) generated by the Fourier transform unit 30 for voice and a short-time spectrum (spectrum including only noise) generated by the Fourier transform unit 31 for noise. ), Stationary noise and non-stationary noise are estimated as described later.
Then, noise removal is performed according to the algorithm of the spectrum subtraction method. The voice data from which noise has been removed by the noise suppression device 32 is output to the voice recognition unit 15.

The noise suppression device 32 is provided with a detection signal (vehicle information) of a vehicle state detection device (vehicle state detection means) 33. The vehicle state detecting device 33 obtains the running state of the automobile via, for example, an ECU, and detects the operating state of each part such as a wiper and a direction indicator by turning on and off a switch (not shown). Further, the vehicle state detection device 33 obtains information such as where the current position of the vehicle is on the map from the control circuit 5 of the car navigation system 1 and the VICS (Vehicle Information and Communica) received by the VICS receiver 34.
It is configured to be able to obtain traffic information of the Action System. Then, the vehicle state detection device 33 outputs these to the noise suppression device 32 as vehicle information.

FIG. 3 is a functional block diagram showing a detailed configuration of the noise suppression device 32. The audio signal data that has been Fourier transformed in the Fourier transform unit 30 is subtracted by a subtractor (first
Stationary noise subtraction means) 35, and
The signal is supplied to a subtractor 35 via a stationary noise estimating section (first stationary noise estimating means) 36 and an open / close switch 37. The noise signal data Fourier-transformed by the Fourier transform unit 31 is directly supplied to the subtractor 38, and is also supplied to the subtractor (the second stationary noise estimating unit) 39 and the open / close switch 40 via the steady-state noise estimator 39. (Second stationary noise subtraction means) 38.

The output data of the subtractor 35 on the voice side is supplied to a subtractor (non-stationary noise subtracting means) 41 and also to a non-stationary noise estimator 42. Also,
The output data of the noise-side subtractor 38 is provided to the non-stationary noise estimator 42. The output data of the non-stationary noise estimating unit (non-stationary noise estimating means) 42 is provided to the subtractor 41 via the open / close switch 43.
Is provided to the speech recognition unit 15.

The opening and closing of the open / close switches 37, 40 and 43 are controlled by a subtraction control section (control means) 44. The subtraction control unit 44 is supplied with voice and noise signal data from the Fourier transform units 30 and 31 and also with vehicle information from the vehicle state detection device 33. The subtraction control unit 44 controls the execution of subtraction in the subtractors 35, 38, and 41 by controlling the opening and closing of the switches 37, 40, and 43 based on the information.

Here, the stationary noise estimating units 36 and 39 are for estimating the stationary noise component contained in the voice signal data and the noise signal data and outputting the data as data. For example, the output data of the noise section given by the Fourier transform unit 30 on the voice side is n1 (t), and the time of the noise section is t1
When 1 → t2, the stationary noise α1 is calculated (estimated) as follows. (Frequency element f is omitted)

(Equation 1) That is, this is the average of the power spectrum in the noise section.

Similarly, assuming that the output data provided by the Fourier transform unit 31 on the noise side is n2 (t), the stationary noise α2 is calculated as follows.

(Equation 2)

Then, if the data of the voice section output by the Fourier transform unit 30 is SN (t), and the output data of the Fourier transform unit 31 is N (t), the subtractor 3
Assuming that the data from which the stationary noise has been removed as a result of the subtraction performed by 5, 5 and 38 are sn (t) and n (t), respectively, sn (t) = SN (t) -α1 (3) n (t) ) = N (t) −α2 (4)

The non-stationary noise estimating section 42 calculates the non-stationary noise r (t) based on the data sn (t) and n (t).
Is estimated. There are various methods for determining the correction coefficient β for estimating the non-stationary noise r (t). For example, assuming that the time of the noise section is t1 → t2 in accordance with the equations (1) and (2), ,

(Equation 3) Can be obtained by Then, the non-stationary noise r (t)
Is represented by r (t) = β · n (t) (6).

Since the non-stationary noise r (t) remains in the audio data sn (t) from which the stationary noise α1 has been removed, the subtractor 41 performs subtraction to perform the non-stationary noise r (t).
Assuming that the audio data from which (t) has been removed is s (t), s (t) = sn (t) -r (t) (7). At this time, if s (t) indicates a negative value,
Clip at an appropriate value (for example, “0”).

That is, since the nature of the stationary noise is constant, it can be relatively easily estimated as an average of the power spectrum. On the other hand, non-stationary noise is noise that suddenly occurs in real time,
Estimation is relatively difficult due to various modes of occurrence, and it is necessary to perform estimation according to a real-time situation and determine whether subtraction is possible or not.

The above is an example in which the non-stationary noise r (t) is obtained on the premise that the stationary noise α1 is removed. When the stationary noise α1 is not removed, the estimation of the non-stationary noise r (t) is performed. Also in the above, estimation is performed using data from which stationary noises α1 and α2 are not removed by opening switches 37 and 40.
That is, the switches 37 and 40 are controlled to open and close in conjunction with each other.

Next, the operation of the present embodiment will be described with reference to FIGS. First, an outline of the operation of the car navigation 1 will be described. When the power of the car navigation 1 is turned on, an operation menu is displayed on the screen of the display device 7.
Then, the driver operates the operation switch group 4 or the remote controller 1
A similar selection instruction is given from the dialog control unit 16 to the control circuit 5 by selecting a process of operating the display unit 4 to display the guidance route on the display device 7 or by inputting a voice through the voice recognition device 9. Then, the processing is executed as follows.

That is, when the driver inputs a destination by voice or the operation of the remote controller 14 with reference to the map displayed on the display device 7, the current position of the vehicle is obtained based on the position data obtained from the GPS receiver 13. Can be And
The control circuit 5 calculates the cost using the Dijkstra method, and obtains the route that is the shortest distance from the current position to the destination as the guidance route. Then, by superimposing and displaying the obtained guidance route on the map of the display device 7, an appropriate route is guided to the driver. The process of calculating the guidance route and the process of providing guidance in this manner are generally well-known, and thus detailed description thereof will be omitted.

Next, the operation of the voice recognition device 9 will be described by taking as an example the case where the destination is input by voice for the above-mentioned route guidance. FIG. 4 is a flowchart showing the control contents of the control unit 23 in the speech recognition device 9. First, the control unit 23 determines whether or not the PTT switch 21 has been pressed (ON) (step S1).
S ", and instructs the amplifier 24 to mute the audio signal output to the audio speaker 25 (step S2).

At that time, if the amplifier 24 is outputting an audio signal to the audio speaker 25, the audio signal is muted as shown in FIGS. 19 and noise microphone 2
The level of the sound signal input to 0 decreases.

Then, the control unit 23 controls the sound extraction device 1
8 is instructed to start the processing for capturing and storing the acoustic signals input to the audio microphone 19 and the noise microphone 20 (step S3). For example, it is assumed that the speaker (driver) utters and inputs the destination as “Showa-cho, Kariya-shi, Aichi”. Then, the frame division unit 26 samples the audio signal input from the microphones 19 and 20 and divides it into frames, and outputs the audio signal data and the noise signal data to the audio section determination device 27. At this time, noise is input to the audio microphone 19 together with the voice of the speaker, and noise is exclusively input to the noise microphone 20.

Then, the voice section determination device 27 calculates, for example, the short-time power for each of the divided frames of the voice signal data and the noise signal data, and determines the frame in which the difference between the two is equal to or more than a predetermined value in the voice input section. Is determined,
The data of the frame is output to the audio buffer 28 and the noise buffer 29, respectively. However, as described above, only one frame located at the head of the buffers 28 and 29 is a noise input section immediately before a voice input section.

Then, the control section 23 waits until the PTT switch 21 is turned off (step S4),
If the answer is F, it is determined to be "YES" and when the voice extraction device 18 is instructed to stop taking in the input signal (Step S5)
Subsequently, an instruction to start the audio extraction processing is given to the audio extraction device 18 (step S6). Then, when the amplifier 24 is instructed to release the mute of the audio signal (step S7), the process proceeds to step S1.

When the speech extraction device 18 is instructed by the control unit 23 to start speech extraction processing, the Fourier transform unit 3
At 0 and 31, frame data corresponding to the voice input section stored in the buffers 28 and 29 are read out, a short-time spectrum is generated by FFT, and output to the noise suppression device 32.

FIG. 6 shows the subtraction control section 4 of the noise suppression device 32.
6 is a flowchart showing control contents of a noise suppression process according to Example 4. The subtraction control unit 44 firstly controls the vehicle state detection device 3
3 (step A1), the respective judgments are made in the following judgment steps A2 to A6, and based on the judgments, the opening and closing of the switches 37, 40 and 43 are controlled, and the noise is reduced by the spectrum subtraction method. Performs suppression suppression.

In the decision step A2, the subtraction controller 4
4 determines whether or not the engine is stopped, and if stopped, determines whether or not the current position of the car is a parking lot based on the navigation information obtained from the control circuit 5 (step). A7). If the current position is the parking lot, the switches 37, 40, and 43 are all opened so that neither the stationary noise nor the non-stationary noise is removed by subtraction (step A8).

That is, when an automobile is parked in a parking lot, engine noise as stationary noise is not generated, and traffic of surrounding vehicles is relatively small, so that non-stationary noise arriving from outside is almost assumed. Not done. Therefore, in such a case, it is possible to obtain audio data with less distortion by not subtracting both the stationary noise and the non-stationary noise. Upon completion of the process in the step A8, the process shifts to the step A1 to obtain the current vehicle information again.

On the other hand, if it is determined in the determination step A2 that the engine is not in the stopped state but "NO", the subtraction control unit 44 determines whether or not the vehicle is stopped (vehicle speed is zero) (step A3). If the vehicle is stopped, the switches 37 and 40 are opened and the switch 43 is closed to perform only the subtraction and removal of the non-stationary noise (step A11).
That is, for example, when the car is stopped at a traffic light at an intersection, the engine sound is low because the engine is idling, and the running sound of the passing car and the operation sound of the turn signal are irregular. These are the dominant situations because they frequently occur as noise having a characteristic. Therefore, by subtracting only the non-stationary noise, the noise can be effectively suppressed without giving distortion as much as possible to the audio data.

When it is determined in the determination step A3 that the vehicle is not in the stopped state but “NO”, the subtraction control unit 44
It is determined whether the vehicle is accelerating or decelerating (step A4). If the vehicle is accelerating or decelerating, the process proceeds to step A11 to perform only subtraction and removal of non-stationary noise. The acceleration / deceleration here is a relatively steep acceleration / deceleration that is equal to or more than a certain value. That is, when the vehicle is accelerating or decelerating, the engine speed changes rapidly, and the engine sound is generated as unsteady noise. Therefore, only non-stationary noise is subtracted and removed.

When the vehicle is not accelerating or decelerating in determination step A4 and determines "NO", the subtraction control unit 44
It is determined whether or not the wiper is operating (step A).
5) If the wiper is operating, the process proceeds to step A11 to perform only the subtraction removal of the non-stationary noise. That is, in the case of rainy weather, the wiper is normally operated, but in such a case, the intermittent operation sound of the rain or the wiper is generated as noise having an unsteady property, and these are dominated by the noise. Is considered relevant. Therefore, only non-stationary noise is subtracted and removed.

If it is determined in the determination step A5 that the wiper is not operating or "NO", the subtraction control unit 44
Based on the VICS information obtained from the VICS receiver 34, it is determined whether or not the traffic volume on the traveling road is equal to or less than a predetermined amount (step A6). If the traffic amount is equal to or less than the predetermined amount, the process proceeds to step A10. Then, by closing the switches 37 and 40 and opening the switch 43, only the subtraction removal of the stationary noise is performed.

That is, if "YES" is determined in the step A6, the vehicle is traveling on a road with a small traffic at a substantially constant speed. Therefore, engine noise, road noise, wind noise and the like are generated in a substantially constant state, while passing noise of other vehicles is hardly generated.
The former, which has stationary characteristics, is dominant. Therefore, by subtracting and removing only the stationary noise, the noise can be effectively suppressed without giving any distortion to the voice data, and the voice recognition rate can be improved.

If the traffic volume exceeds the predetermined volume in the determination step A6 and the determination is "NO", the subtraction control unit 44 proceeds to step A10. Then, by closing the switches 37, 40 and 43, the subtraction and removal of the stationary noise and the non-stationary noise are performed. That is, in the determination step A6, "N
If it is determined to be "O", the vehicle is running at a substantially constant speed on a road with a certain amount of traffic. Therefore, engine noise, road noise, and the like are generated in a substantially constant state, and passing noise of another vehicle is also generated at a certain frequency. Therefore, both the stationary noise and the non-stationary noise are subtracted and removed.

As described above, when noise suppression is removed by the noise suppression device 32, the voice data s (t) is output to the voice recognition unit 15. Then, as described above, the speech recognition unit 15 obtains time-series data in which acoustic features are extracted from the speech data s (t), and the time-series data is divided into sections, and the dictionary is divided into appropriate words. Collation with the data pattern registered in the data is performed.

The recognition result is provided from the voice recognition unit 15 to the dialogue control unit 16, and the dialogue control unit 16 gives an instruction to utter a response voice to the voice synthesis unit 17 based on the provided recognition result. Through the speaker 22, a synthesized voice is uttered with "Showa-cho, Kariya city, Aichi prefecture". Also, the voice recognition unit 15
Is also given to the control circuit 5, and the control circuit 5
Starts the calculation of the guide route with "Showa-cho, Kariya-shi, Aichi" as the destination.

As described above, according to the present embodiment, the subtraction control unit 44 of the noise suppression device 32 controls the opening and closing of the switches 37, 40 and 43 based on the vehicle information obtained from the vehicle information detection device 33. Thus, the subtraction and removal of the stationary noise and the non-stationary noise from the voice signal data of the speaker input to the voice microphone 19 are performed according to the running state of the automobile.

That is, distortion is given to voice data by performing noise suppression removal so as to be adapted to the state of noise generated according to the running state of the automobile (for example, as described above, s (t ) Can be suppressed as much as possible, and the recognition rate in the speech recognition device 9 can be improved. Since the car navigation system 1 performs a predetermined process based on the voice data pattern recognized by the voice recognition device 9, the voice of the driver as a speaker can be accurately detected even in a vehicle room where the noise environment fluctuates drastically. Can be recognized. Therefore, the driver can reliably perform a desired operation by voice input without manually operating the operation switch group 4 and the remote controller 14, thereby improving convenience and ensuring driving safety. Can also.

Further, according to the present embodiment, the subtraction control unit 44 of the noise suppression device is specifically configured to control the state in which the vehicle is parked in a parking lot, the state in which the vehicle is stopped at an intersection, and the operation of the wiper. The steady noise and the non-stationary noise are subtracted and removed appropriately according to the state, the acceleration / deceleration state of the vehicle, the traffic volume, and the like. By performing the suppression removal, the recognition rate of the speech recognition device 9 can be increased.

(Second Embodiment) FIG. 7 shows a second embodiment of the present invention. The same parts as those in the first embodiment are denoted by the same reference numerals, and description thereof will be omitted. I will explain only. In FIG. 7 showing a detailed configuration of the noise suppression device 45, audio signal data from the Fourier transformer 30 is
The noise signal data from the Fourier transform unit 31 is provided to a subtractor (unsteady noise subtracting means) 46, a non-stationary noise estimating unit (unsteady noise estimating means) 47 and a subtraction control unit (control means) 48. It is provided to the non-stationary noise estimation unit 47 and the subtraction control unit 48.

The output data of the subtractor 46 is supplied to the speech recognition section 15 via a subtractor (stationary noise subtracting means) 49 and is also supplied to a stationary noise estimating section (stationary noise estimating means) 5.
The signal is given to the subtractor 49 via the “0” and the open / close switch 51. The output data of the non-stationary noise estimation unit 47 is
The signal is provided to the subtractor 46 via the open / close switch 52. The opening and closing of the open / close switches 51 and 52 is controlled by the subtraction control unit 48. Other configurations are the same as in the first embodiment.

Non-stationary noise estimating section 47 in the second embodiment
Is designed to estimate non-stationary noise from speech signal data and noise signal data supplied from the Fourier transform units 30 and 31 without suppressing the stationary noise. That is, a correction coefficient β for estimating the unsteady noise r (t) is, for example,

(Equation 4) And the non-stationary noise r (t) is determined by r (t) = β · N (t) (9) Therefore, if the output data of the subtractor 46 when the switch 52 is closed is sn1 (t), then sn1 (t) = SN (t) -r (t) (10)

Further, the stationary noise estimating section 50 obtains the stationary noise α1 similarly to the stationary noise estimating section 36 in the first embodiment. Therefore, assuming that the output data of the subtractor 49 when the switch 51 is closed is sn2 (t), then sn2 (t) = sn1 (t) -α1 (11). That is, the output data of the subtractor 49 when the switch 51 is closed and the switch 52 is opened is the same as the equation (3).

Next, the operation of the second embodiment will be described. The subtraction control unit 48 performs the noise suppression process according to the flowchart shown in FIG. 6 similarly to the subtraction control unit 44 of the first embodiment, but performs steps A8 to A related to noise suppression removal.
The open / close state of the switch in 11 is different. That is, switch 51 switch 52 Step A8: Open without removing noise Step A9: Remove stationary and non-stationary noise Close Close Step A10: Remove only stationary noise Close Open Step A11: Remove only non-stationary noise Open Close .

According to the second embodiment configured as described above, when the non-stationary noise estimating unit 47 estimates the non-stationary noise, the non-stationary noise is not suppressed and removed from the voice signal data and the noise signal data. When the noise is suppressed and removed, the noise is collectively removed at the output side of the subtractor 46, so that the configuration is simplified and the arithmetic processing can be reduced.

(Third Embodiment) FIGS. 8 and 9 show a third embodiment of the present invention. The same parts as those of the first embodiment are denoted by the same reference numerals, and the description thereof will be omitted. Only the parts will be described. In FIG. 8 showing the detailed configuration of the noise suppression device 53, the audio signal data from the Fourier transform unit 30 is supplied to a stationary noise subtraction unit (first stationary noise subtraction unit) 54 and a subtractor (nonstationary noise subtraction unit) 55. The signal is provided to the voice recognition unit 15 via the control unit and also to a subtraction control unit (control means) 56. Further, the Fourier transform unit 31
Is supplied to a stationary noise subtraction section (second stationary noise subtraction means) 57 and a subtraction control section 56, respectively.

The stationary noise subtracting sections 54 and 57 estimate the stationary noise of the voice signal data and the noise signal data in the same manner as the stationary noise estimating sections 36 and 39 in the first embodiment, and perform the subtraction processing of the estimated stationary noise data. Until you do. The output data of the stationary noise subtracting units 54 and 57 are respectively supplied to a non-stationary noise estimating unit (non-stationary noise estimating unit) 58, and the output data of the non-stationary noise estimating unit 58 is transmitted via an open / close switch 59. To the subtractor 55. Note that the non-stationary noise estimation unit 58 also
The non-stationary noise is estimated similarly to the non-stationary noise estimating unit 42 in the first embodiment. The opening and closing of the open / close switch 59 is controlled by the subtraction control unit 56. Other configurations are the same as in the first embodiment.

Next, the operation of the third embodiment will be described with reference to FIG. In the configuration of the third embodiment, the stationary noise is always subtracted by the stationary noise subtractors 54 and 57, and the subtraction controller 56 determines only whether or not to perform the nonstationary noise subtraction. To open and close the switch 59. That is, in the flowchart shown in FIG.
Steps A2 to A5, A7, A8
Is subtracted, and when only the traffic information is obtained as the vehicle information, the subtraction controller 56 closes the switch 59 and suppresses and removes both the stationary and non-stationary noises based on the information (step A9). ) Or whether the switch 59 is opened to suppress and remove only the stationary noise (step A10).

According to the third embodiment configured as described above, the subtraction control unit 56 determines whether to suppress and remove the non-stationary noise based only on the traffic information. That is, even if the surrounding situation changes, an appropriate value can be estimated in accordance with the situation (because it is obtained as an average value). It is also assumed that data will not be greatly distorted. Therefore, with the configuration as in the third embodiment, the subtraction control can be performed simply. Further, information obtained from the vehicle state detection device 33 can be reduced (in this case, only the information of the VICS receiver 34 is sufficient).

The present invention is not limited to the embodiment described above and shown in the drawings, and the following modifications or extensions are possible. The flowcharts shown in FIGS. 6 and 9 are merely examples, and the factors for determining whether stationary or non-stationary noise removal is possible are added, changed, and changed as appropriate according to the situation.
Settings such as deletion can be made. For example, when the present invention is applied to a device used in a factory or the like other than the car navigation system 1 that is operated by voice input, conditions may be set according to the type and number of moving machine tools. For example, when a device that generates non-stationary noise such as a press machine is operating, only the non-stationary noise is removed, and noise having a non-stationary characteristic such as a lathe or drilling machine is generated. When the apparatus for operating is operating, the setting is made to remove only the stationary noise.

The speech recognition device 9, the speech extraction device 18, and the noise suppression device 32 may be independently constructed. For example, weather information (for example, rainfall and wind speed)
May be directly input, and the noise suppression device may determine whether or not to remove noise according to the weather condition. The subtraction control units 44, 48, and 56 analyze the voice signal data and the noise signal data supplied from the Fourier transform units 30 and 31 to directly estimate the noise generation state, and determine whether stationary or non-stationary noise can be removed. The determination may be made. The information of the vehicle state detecting means may be input to the non-stationary noise estimating unit, and the calculation method of the correction coefficient β for estimating the non-stationary noise may be changed according to the vehicle information.
Further, it is not always necessary to use the correction coefficient β, and the correction coefficient β may simply be obtained by removing the stationary noise component as the non-stationary noise.

The noise suppression device and the voice recognition device are not limited to the vehicular navigation device, but may be applied to any device that performs some processing based on the voice recognition result.

[Brief description of the drawings]

FIG. 1 illustrates a first embodiment in which the present invention is applied to a car navigation device, and is a functional block diagram illustrating an electrical configuration of the entire car navigation device.

FIG. 2 is a functional block diagram showing a detailed configuration of a voice extraction device.

FIG. 3 is a functional block diagram showing a detailed configuration of the noise suppression device.

FIG. 4 is a flowchart showing control contents of a control unit of the voice recognition device.

5A and 5B are waveform diagrams, respectively, wherein FIG. 5A shows an ON / OFF state of a PTT switch, FIG. 5B shows an input signal of an audio microphone,
(C) shows an input signal of the noise microphone.

FIG. 6 is a flowchart showing control contents of a subtraction control unit of the noise suppression device.

FIG. 7 is a view corresponding to FIG. 3, showing a second embodiment of the present invention;

FIG. 8 is a view corresponding to FIG. 3, showing a third embodiment of the present invention.

FIG. 9 is a diagram corresponding to FIG. 6;

[Explanation of symbols]

1 is a car navigation device (vehicle navigation device), 9 is a voice recognition device, 15 is a voice recognition unit (voice recognition means), 19 is a voice microphone, 20 is a noise microphone, 3
2 is a noise suppression device, 33 is a vehicle state detection device (vehicle state detection means), 34 is a VICS receiver, 35 is a subtractor (first steady noise subtraction means), 36 is a steady noise estimator (first steady noise estimation means). ), 38 are subtractors (second stationary noise subtracting means), 39 is a stationary noise estimating section (second stationary noise estimating means), 41 is a subtractor (nonstationary noise subtracting means), and 42 is a nonstationary noise estimating section ( Non-stationary noise estimation means), 44 is a subtraction control unit (control means), 45 is a noise suppression device, 46 is a subtractor (non-stationary noise subtraction means), 47 is a non-stationary noise estimation unit (non-stationary noise estimation means), 48 is a subtraction control unit (control means);
50 is a stationary noise estimating unit (stationary noise estimating means), 49 is a subtractor (stationary noise subtracting means), 53 is a noise suppressing device, 54
Is a stationary noise subtraction unit (first stationary noise subtraction unit), 55 is a subtractor (nonstationary noise subtraction unit), 56 is a subtraction control unit (control unit), and 57 is a stationary noise subtraction unit (second stationary noise subtraction unit). And 58 are non-stationary noise estimating units (non-stationary noise estimating means)
Is shown.

────────────────────────────────────────────────── ───

[Procedure amendment]

[Submission date] May 21, 1999 (1999.5.2
1)

[Procedure amendment 1]

[Document name to be amended] Statement

[Correction target item name] 0025

[Correction method] Change

[Correction contents]

[0025]

DESCRIPTION OF THE PREFERRED EMBODIMENTS (First Embodiment) Hereinafter, a first embodiment in which the present invention is applied to a car navigation apparatus will be described with reference to FIGS. In FIG. 1 showing the electric configuration of the entire car navigation device 1, the car navigation device 1 includes a position detector 2, a map data input device 3, an operation switch group 4, a control circuit 5 to which these are connected, and the like. . Further, an external memory 6, a display device 7 such as a display, a remote control sensor 8, a voice recognition device 9, and the like are connected to the control circuit 5. The control circuit 5 is configured as a microcomputer,
Although not specifically shown, a well-known CP
U, ROM, RAM, I / O, etc. These components are electrically connected to each other via a bus line.

 ──────────────────────────────────────────────────続 き Continued on the front page F term (reference) 2F029 AA02 AB01 AB07 AB13 AC02 AC08 AC09 AC14 AC18 AD01 5D015 DD02 EE05 KK01 5H180 AA01 BB12 BB13 CC11 CC30 EE02 FF04 FF05 FF07 FF12 FF22 FF25 FF27

Claims (14)

[Claims] An audio microphone for inputting audio, a noise microphone for inputting noise, and a stationary noise estimated based on an input signal of the audio microphone are subtracted from the input signal. 1 stationary noise subtracting means, second stationary noise subtracting means for subtracting the stationary noise estimated based on the input signal of the noise microphone from the input signal, output signal of the first stationary noise subtracting means, and (2) a non-stationary noise estimating means for estimating non-stationary noise based on an output signal of the stationary noise subtracting means; and a non-stationary noise estimated by the non-stationary noise estimating means from an output signal of the first stationary noise subtracting means. A noise suppression device comprising: a non-stationary noise subtracting unit that performs subtraction; and a control unit that controls execution of subtraction in the non-stationary noise subtracting unit according to a surrounding state. 2. An audio microphone for inputting audio, a noise microphone for inputting noise, first stationary noise estimating means for estimating stationary noise based on an input signal of the audio microphone, Second stationary noise estimating means for estimating stationary noise based on the input signal of the noise microphone; and first subtracting the stationary noise estimated by the first stationary noise estimating means from the input signal of the voice microphone. Stationary noise subtracting means; second stationary noise subtracting means for subtracting the stationary noise estimated by the second stationary noise estimating means from the input signal of the noise microphone; input signal of the audio microphone and the noise A non-stationary noise estimating unit for estimating the non-stationary noise based on an input signal of the microphone for use, and the non-stationary noise estimating unit estimating from an output signal of the first stationary noise subtracting unit. Non-stationary noise subtracting means for subtracting non-stationary noise; and control means for controlling execution of subtraction in the first and second stationary noise subtracting means and the non-stationary noise subtracting means, respectively, according to the surrounding state. A noise suppressor characterized by: 3. An audio microphone for inputting audio, a noise microphone for inputting noise, and estimating non-stationary noise based on an input signal of the audio microphone and an input signal of the noise microphone. Non-stationary noise estimating means for performing, stationary noise estimating means for estimating stationary noise based on the input signal of the audio microphone, and non-stationary noise subtracting means for subtracting the non-stationary noise from the input signal of the audio microphone. A stationary noise subtracting unit for subtracting the stationary noise estimated by the stationary noise estimating unit from an output signal of the nonstationary noise subtracting unit; and the nonstationary noise subtracting unit and the stationary noise subtracting unit according to a surrounding state. And a control means for controlling each execution of the subtraction in the noise suppression device. 4. A speech recognition apparatus comprising: the noise suppression apparatus according to claim 1; and speech recognition means for analyzing speech data output by the noise suppression apparatus and recognizing a speech data pattern. apparatus. 5. A noise suppressor according to claim 2 or 3, further comprising: voice recognition means for analyzing voice data output by said noise suppressor and recognizing a voice data pattern. Voice recognition device. 6. A vehicular navigation device comprising the voice recognition device according to claim 5, wherein a predetermined process is executed based on a voice data pattern recognized by the voice recognition means. 7. A vehicle state detection means for detecting a state of a vehicle and outputting a detection signal to the voice recognition device, wherein the control means of the voice recognition device outputs a detection signal to the detection signal output from the vehicle state detection means. 7. The vehicular navigation device according to claim 6, wherein the execution of the subtraction in the stationary noise subtracting means and the non-stationary noise subtracting means is controlled accordingly. 8. The control means of the noise suppression device controls so that only the subtraction in the non-stationary noise subtraction means is performed when a wiper of the vehicle is operating. 8. The vehicle navigation device according to claim 7. 9. The noise suppression device according to claim 7, wherein, when the vehicle is stopped, the control unit controls the non-stationary noise subtraction unit to perform only the subtraction. 9. The vehicle navigation device according to claim 8. 10. The control means of the noise suppression device controls so that only the subtraction in the non-stationary noise subtraction means is executed when the vehicle is accelerating or decelerating. Item 10. The vehicle navigation device according to any one of Items 7 to 9. 11. The vehicle state detecting means is configured to be able to receive also navigation information of a vehicle, and the control means of the noise suppression device, according to navigation information obtained via the vehicle state detecting means, 11. The vehicular navigation apparatus according to claim 7, wherein execution of subtraction in said stationary noise subtracting means and non-stationary noise subtracting means is controlled. 12. The control means of the noise suppression device, when the vehicle is located in a parking lot and the engine of the vehicle is stopped, the subtraction by the stationary noise subtraction means and the non-stationary noise subtraction means. 12. The vehicular navigation device according to claim 11, wherein control is performed so that neither of them is executed. 13. The vehicle state detecting means includes a receiver for receiving VICS information, and the control means of the noise suppression device, according to traffic information obtained from the receiver via the vehicle state detecting means, 13. The vehicular navigation device according to claim 7, wherein execution of subtraction in the stationary noise subtracting unit and the non-stationary noise subtracting unit is controlled. 14. The control means of the noise suppression device, when the vehicle is traveling at a substantially constant speed and the traffic on the road on which the vehicle is currently traveling is equal to or less than a predetermined amount,
14. The navigation apparatus for a vehicle according to claim 13, wherein control is performed such that only the subtraction in the stationary noise subtraction unit is performed.