Multi-channel speech processing device, hearing aid system and speech processing method

Abstract

The present invention discloses a multi-channel speech processing device, a hearing aid system and a speech processing method, which can meet the audio processing requirements of different frequency bands and improve the speech signal processing effect. The multi-channel speech processing device uses a high-pass filter, an analysis filter, an FFT unit, a noise elimination unit, a hearing compensation unit, a frequency equalization unit and an IFFT unit to respectively perform high-pass filtering, analysis, fast Fourier transform, noise elimination processing, hearing compensation, frequency equalization processing and inverse Fourier transform on a digital speech signal; the hearing aid system includes a speech signal acquisition unit, a main control state machine FSM, a multi-channel speech processing device, an analog circuit, a power supply module and a speech signal output unit, which amplifies and converts the original speech signal in sequence through a PGA amplifier circuit and an SD_ADC analog-to-digital conversion circuit, and processes the digital modulated signal through the multi-channel speech processing device to obtain a reconstructed speech signal.

Description

Technical Field

[0001] The present invention relates to the technical field of speech signal processing for hearing aid applications, and in particular to a multi-channel speech processing device, a hearing aid system, and a speech processing method. Background Art

[0002] Hearing aids are assistive devices used by the hearing-impaired. As sound amplification devices, they effectively amplify external sounds, helping those with hearing impairments hear them, thereby improving their hearing ability and enhancing their quality of life. With the development of informatization and the widespread use of electronics, the hearing-impaired are placing increasingly high demands on the speech signal processing performance of hearing aids.

[0003] Currently, the vast majority of hearing aids on the market are analog hearing aids. Analog hearing aids process speech signals as analog signals and linearly amplify them through analog amplifiers. However, analog hearing aids primarily use an overall amplification method to process full-band audio signals. This method cannot effectively filter and eliminate external noise, resulting in unclear hearing of quiet sounds and hearing damage caused by loud sounds, seriously affecting the wearing and use experience of hearing-impaired people. In addition, among the hearing-impaired population, the degree and status of hearing loss vary among different hearing-impaired patients, and most of them experience different degrees of hearing loss in different frequency bands. The above-mentioned full-band audio processing method cannot meet the hearing compensation needs of different hearing patients. In addition, hearing aids generally need to be worn continuously for a long time. Long-term wearing of such analog hearing aids will also gradually damage other intact hearing. Summary of the Invention

[0004] In view of the above-mentioned deficiencies in the prior art, the present invention provides a multi-channel voice processing device, which can meet the requirements of audio processing in different frequency bands and improve the voice signal processing effect.

[0005] To achieve the above object, the present invention adopts the following technical solutions:

[0006] A multi-channel speech processing device includes a speech signal input unit, which inputs a digital modulation signal. The device is characterized in that the speech signal input terminal is sequentially connected to a modulator DEM, a high-pass filter HPF, an analysis filter bank WOLA_A, an FFT unit, a noise elimination unit NR, a hearing compensation unit WDRC, a frequency equalization unit EQ, an IFFT unit, a synthesis filter bank WOLA_S, a low-pass filter LPF, and a volume output unit VOLUME.

[0007] The modulator DEM is used to restore the digital modulated signal to a digital voice signal, and the high-pass filter HPF, the analysis filter bank WOLA_A, the FFT unit, the noise elimination unit, the hearing compensation unit, the frequency equalization unit, and the IFFT unit are used to perform high-pass filtering, analysis, fast Fourier transform, noise elimination processing, hearing compensation, frequency equalization processing, and inverse Fourier transform on the digital voice signal, respectively, to obtain a reconstructed speech signal;

[0008] The volume output unit VOLUME is used to output the reconstructed language signal.

[0009] It is further characterized in that

[0010] The analysis filter bank WOLA_A comprises at least one first filter bank connected in parallel, wherein the first filter bank divides the speech signal processed by the high-pass filter HPF into a corresponding number of sub-bands according to the frequency segment definition;

[0011] The integrated filter bank WOLA_S comprises at least one second filter bank connected in parallel, wherein the second filter bank is used to add the speech signal after the fast Fourier transform of the FFT unit;

[0012] The first filter group includes K channels: the 0th channel to the K-1th channel, the 0th channel is a low-pass filter, the K-1th channel is a high-pass filter, and the remaining channels are band-pass filters, that is, the second channel to the K-2th channel are band-pass filters.

[0013] A method for processing a voice signal using the multi-channel voice processing device is characterized in that the method comprises: A1, inputting a digital modulation signal through a voice signal input terminal;

[0014] A2. The digital modulated signal is restored to a digital voice signal through the modulator DEM;

[0015] A3. The digital speech signal is subjected to high-pass filtering, analysis, fast Fourier transform (FFT) to convert the time domain signal into a frequency domain signal, noise elimination, hearing compensation, frequency equalization, and inverse Fourier transform (IFFT) to convert the frequency domain signal into a time domain signal, respectively, through a high-pass filter (HPF), an analysis filter bank (WOLA_A), an FFT unit, a noise cancellation unit, a hearing compensation unit, a frequency equalization unit, and an IFFT unit, to obtain a reconstructed speech signal.

[0016] A4. Output the reconstructed speech signal through a volume output unit VOLUME.

[0017] It is further characterized in that

[0018] In step A3, the input second speech signal is separated by the first filter group in the analysis filter group WOLA_A, and the second speech signal is evenly divided into several frequency band signals. The number of frequency band signals is consistent with the number of channels. After the signal separation, each frequency band signal is sequentially operated, noise eliminated, hearing compensated, and frequency equalized in the corresponding channel.

[0019] A hearing aid system is encapsulated in a hearing aid housing, the hearing aid system comprising a voice signal acquisition unit, a main control state machine (FSM), an analog circuit, a power module, and a voice signal output unit. The main control state machine (FSM) is used to control the hearing aid system. The analog circuit comprises a PGA amplifier circuit, an SD_ADC analog-to-digital conversion circuit, and an audio power amplifier (PA). The voice signal acquisition unit is sequentially connected to the PGA amplifier circuit and the SD_ADC analog-to-digital conversion circuit. The voice signal acquisition unit is used to acquire an original voice signal. The PGA amplifier circuit is used to amplify the original voice signal. The SD_ADC analog-to-digital conversion circuit is used to perform analog-to-digital conversion on the voice signal amplified by the PGA amplifier circuit. The power module is used to power the hearing aid system. The hearing aid system includes the above-mentioned multi-channel voice processing device, the input end of the multi-channel voice processing device is connected to the SD_ADC analog-to-digital conversion circuit, and the output end of the multi-channel voice processing device is sequentially connected to the audio power amplifier (PA) and the voice signal output unit.

[0020] The audio power amplifier PA is used to amplify the power of the reconstructed speech signal obtained after processing by the multi-channel speech processing device;

[0021] The speech signal output unit is used to output the amplified and reconstructed speech signal.

[0022] It is further characterized in that

[0023] The master state machine FSM includes an I2C unit, a key detection unit KEYDET, a volume and prompt tone control unit CTRL, and a digital module control unit DMCU. The I2C unit is used to communicate with an off-chip storage unit and directly access internal registers through an off-chip debugging communication interface. The key detection unit KEYDET is connected to a key KEY and is used to detect a switch signal input by the key KEY. The volume and prompt tone control unit CTRL is used to control the volume of the hearing aid and to superimpose a prompt tone in the reconstructed speech signal. The digital module control unit DMCU is used to convert off-chip debugging control instructions into internal read-write control signals, and to convert the switch signal input by the key KEY into corresponding parameter read-write control signals.

[0024] The voice signal acquisition unit includes a microphone MIC;

[0025] The voice signal output unit includes a speaker SPEAK;

[0026] The analog circuit further includes a sigma-delta modulator and a power amplifier unit PA. The sigma-delta modulator is used to convert the reconstructed speech signal into a 1-bit modulation signal. The power amplifier unit PA is used to amplify the 1-bit modulation signal to obtain an amplified speech signal.

[0027] The method for processing a speech signal based on the above-mentioned hearing aid system is characterized in that the method comprises: B1, collecting an original speech signal by a speech signal acquisition unit and sending it to a PGA amplification circuit in an analog circuit, amplifying and performing analog-to-digital conversion on the original speech signal in sequence by the PGA amplification circuit and the SD_ADC analog-to-digital conversion circuit to obtain a digital modulation signal, and sending the digital modulation signal to a multi-channel speech processing device;

[0028] B2. Processing the digital modulated signal through a multi-channel speech processing device to obtain a reconstructed speech signal;

[0029] B3. Output the reconstructed language signal through the voice signal output unit.

[0030] The above-mentioned method of the present invention can achieve the following beneficial effects: the high-pass filter HPF, the noise elimination unit NR, and the low-pass filter LPF in the multi-channel speech processing device are used to filter and eliminate external noise mixed in the digital modulated signal, which is conducive to improving the speech signal processing effect and obtaining a clearer and more stable speech signal. In addition, the multi-channel speech processing device is also provided with an analysis filter group WOLA_A, a synthesis filter group WOLA_S, a hearing compensation unit WDRC, and a frequency equalization unit EQ. The analysis filter WOLA_A is used to divide the digital speech signal into different frequency band signals, which is conducive to compensating the different frequency band signals separately, thereby meeting the hearing compensation needs of patients with different hearing impairments. The noise-reduced speech signal is synthesized by the synthesis filter group WOLA_S, and the synthesized speech signal is compensated and equalized by the hearing compensation unit WDRC and the frequency equalization unit EQ, which is conducive to obtaining a clearer and more balanced reconstructed speech signal, thereby greatly improving the speech signal restoration effect.

[0031] The above-mentioned multi-channel speech processing device is applied to the hearing aid system. The digital modulated signal is divided into several frequency band signals of different frequency bands (different channels) through the multi-channel speech processing device, which is conducive to realizing wide dynamic compression hearing compensation, frequency equalization and other refined processing of different frequency band signals. This frequency band audio processing method of amplifying different frequency band signals can realize language signal reconstruction even under different allocation requirements of people with different hearing loss. The reconstructed language speech signal obtained after multi-channel refined processing is closer to the original speech signal, thereby avoiding the problem of unclear hearing due to soft voice and hearing damage caused by loud voice, and improving the wearing and using experience of hearing aids for people with hearing impairment. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. The drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0033] Figure 1 is a structural block diagram of the hearing aid system of the present invention;

[0034] Figure 2 A flow chart of processing a speech signal using the multi-channel speech processing device of the present invention;

[0035] Figure 3 A specific flow chart of processing a speech signal using the multi-channel speech processing device of the present invention;

[0036] Figure 4 It is a time domain effect diagram of the original speech signal and the reconstructed speech signal after the speech signal is processed by the speech processing device of the present invention;

[0037] Figure 5 It is a frequency domain effect diagram of the original speech signal and the reconstructed speech signal after the speech signal is processed by the speech processing device of the present invention. DETAILED DESCRIPTION

[0038] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0039] It should be noted that the terms "including" and "having" and any variations thereof in the specification and claims of the present invention and the above-mentioned drawings are intended to cover non-exclusive inclusions. For example, a process, method, apparatus, product or equipment that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or are inherent to these processes, methods, products or equipment.

[0040] To address the technical problem that analog hearing aids in the prior art use overall amplification to process full-band audio, resulting in poor voice signal processing and an inability to meet the requirements for audio processing in different frequency bands, a specific embodiment of a hearing aid system is provided below.

[0041] See Figure 1 A hearing aid system is encapsulated in a hearing aid shell. The hearing aid system includes a voice signal acquisition unit 1, a main control state machine FSM, an analog circuit ALG, a power module (not shown in the figure), and a voice signal output unit 2. The voice signal acquisition unit includes a microphone MIC, which collects original voice signals through the microphone. The main control state machine FSM is used to control the hearing aid system. The main control state machine FSM includes an I2C unit, a key detection unit KEYDET, a volume and prompt tone control unit CTRL, and a digital module control unit DMCU. The I2C unit has two functions. The first function is to communicate with the off-chip storage unit EEPROM, and the second function is to use the off-chip debugging communication interface to directly access the internal registers. The key detection unit KEYDET is connected to the key KEY. The key detection unit KEYDET is used to detect the switch signal input by the key KEY. The volume and prompt tone control unit CTRL is used to control the volume of the hearing aid and to superimpose the prompt tone in the reconstructed language signal; the digital module control unit DMCU has the first function of converting the off-chip I2C debugging control instructions into internal read and write control signals, and the second function is to convert the switch signal input by the key KEY into the corresponding parameter read and write control signals.

[0042] The analog circuit includes a PGA amplifier circuit, an SD_ADC analog-to-digital conversion circuit, an audio power amplifier (PA), a sigma-delta modulator, and a power amplifier unit (PA). The voice signal acquisition unit is sequentially connected to the PGA amplifier circuit and the SD_ADC analog-to-digital conversion circuit. The PGA amplifier circuit is used to amplify the original voice signal; the SD_ADC analog-to-digital conversion circuit is used to perform analog-to-digital conversion on the voice signal amplified by the PGA amplifier circuit; the audio power amplifier (PA) is used to amplify the power of the reconstructed speech signal obtained after processing by the multi-channel speech processing device to obtain an amplified reconstructed speech signal; the sigma-delta modulator is used to convert the reconstructed speech signal into a 1-bit modulated signal; the power amplifier unit (PA) is used to amplify the 1-bit modulated signal to obtain an amplified voice signal. The power module is used to power the hearing aid system. The voice signal output unit includes a speaker (SPEAK) for outputting the amplified reconstructed speech signal.

[0043] The hearing aid system is also provided with a multi-channel speech processing device MC_DSM. The input end of the speech signal input unit in the multi-channel speech processing device is connected to the output end of the SD_ADC analog-to-digital conversion circuit. The output end of the multi-channel speech processing device is sequentially connected to the audio power amplifier PA and the speech signal output unit.

[0044] The specific connection structure of the multi-channel speech processing device is as follows: the speech signal input terminal is sequentially connected to a modulator DEM, a high-pass filter HPF, an analysis filter bank WOLA_A, an FFT unit, a noise cancellation unit NR, a hearing compensation unit WDRC, a frequency equalization unit EQ, an IFFT unit, a synthesis filter bank WOLA_S, a low-pass filter LPF, and a volume output unit VOLUME. In this embodiment, the analysis filter WOLA_A includes a set of first filter banks connected in parallel. The frequency response characteristics of the first filter bank are such that the signal is divided into a corresponding number of sub-bands. Specifically, the first filter bank is a group of filters having a common input signal or a common output signal. In this embodiment, the first filter bank is composed of a set of system functions {H k The integrated filter bank WOLA_S includes a set of second filter banks connected in parallel. The second filter bank is used to add the speech signal after the Fourier transform of the FFT unit. The second filter bank is composed of a set of system functions {G k (z)}, the corresponding output is {y k (n)} filter banks are connected in parallel, and the outputs of each filter are added together to form a signal

[0045] The specific functions of each unit are as follows: the modulator DEM is used to restore the digital modulated signal to a digital voice signal, the high-pass filter HPF is used to perform high-pass filtering on the digital voice signal, the analysis filter group WOLA_A is used to analyze the signal filtered by the high-pass filter HPF, the FFT unit is used to operate on the signals of each frequency band divided by the analysis filter group WOLA_A and convert them into the frequency domain, the noise elimination unit NR is used to perform noise elimination processing on the signal after the operation of the FFT unit, the hearing compensation unit WDRC is used to perform hearing compensation on the signal processed by the noise elimination unit NR, the frequency equalization unit EQ is used to perform frequency equalization processing on the signal compensated by the hearing compensation unit WDRC, the integrated filter group WOLA_S is used to integrate the signal equalized by the frequency equalization unit EQ, the IFFT unit is used to perform inverse Fourier transform on the signal after the EQ equalization processing to convert the signal into the time domain, the low-pass filter LPF is used to low-pass filter the signal integrated by the integrated filter group WOLA_S to obtain the reconstructed language signal; the volume output unit VOLUME is used to output the reconstructed language signal.

[0046] The multi-channel voice processing device is used to process voice signals. The specific processing steps include: A1, inputting a digital modulation signal through the voice signal input terminal;

[0047] A2. The digital modulated signal is restored to a digital voice signal through the modulator DEM;

[0048] A3. The digital speech signal is subjected to high-pass filtering, analysis, fast Fourier transform to convert the time domain signal into a frequency domain signal, noise elimination processing, hearing compensation, frequency equalization processing, and inverse Fourier transform to convert the frequency domain signal into a time domain signal through the high-pass filter HPF, analysis filter group WOLA_A, FFT unit, noise elimination unit NR, hearing compensation unit WDRC, frequency equalization unit EQ, and IFFT unit in sequence to obtain a reconstructed speech signal. Among them, the first filter group in the analysis filter group WOLA_A is mainly used to perform multi-channel separation on the signal filtered by the high-pass filter HPF, that is, to evenly divide the signal filtered by the high-pass filter HPF into multiple frequency band signals, and then perform Fourier transform, noise elimination processing, hearing compensation, frequency equalization, inverse Fourier transform and other related signal processing on each frequency band signal separately, and then the reconstructed speech signals of each channel are synthesized into one signal through the comprehensive filter group WOLA_S. Specifically, see Figure 2 , each of the K channels in the first filter group has a filter h k (n), k=0,1,...,K-1, K is an integer. After the input signal x(n) enters K channels, it is divided into K sub-band signals X by the filter in each channel.k (m); sub-band signal X k (m) are processed separately and become Every way The signal passes through the corresponding finite impulse response (FIR) filter g again k (n) Filtering process to obtain Finally, add them together to form a signal Output, where and x(n) are broadband signals, and the intermediate signals in the processing are subband signals; and x(n),X k (m) and The values ​​of n are not necessarily the same, which depends on the intermediate signal processing process. In this embodiment, n is 32 or 64.

[0049] {H k (z), k = 0, 1, 2, ... K-1} is derived from the prototype filter H0(z), which is generally an FIR filter or an IIR filter:

[0050]

[0051] Where K represents the number of channels, N is an integer, and z represents the sampling value of the input speech signal in the frequency domain;

[0052] Therefore, the filter {H k (z), k=0,1,2,…K-1} can be simply obtained by The frequency response of the prototype filter is obtained by uniformly shifting the frequency response by multiples of . The impulse response of the filter in the time domain is:

[0053]

[0054] h0(n) is the impulse response of the prototype filter, which is usually an FIR filter or an IIR filter. If H0(z) is the system function of the prototype filter, then the system function of the kth filter is:

[0055]

[0056] Since the signal in each channel during processing For a sub-band signal, the sampling rate of each channel can be reduced. For a filter bank with uniformly separated frequency bands, if the signal x(n) is full-band (i.e., the bandwidth is 2π), after being divided by K filters, the width of each frequency band is Then theoretically, the sampling rate of each channel signal can be reduced by a factor M: M<=K. When M=K, it is critical sampling, where M is the sampling ratio. The sampling ratio M is determined by the design of the filters h0(n) and g0(n). h0(n) is the impulse response of the prototype filter, generally an FIR filter or an IIR filter, and the synthesis window g0(n) should also be a finite impulse response (FIR) filter.

[0057] See Figure 3 , by using complex exponential Multiply by x(n) to get x(n) The high-frequency spectrum of x(n) is shifted to low frequency, and the result is passed through a low-pass filter with an impulse response of h0(n), which is abbreviated as h(n) below. The high-frequency components in the signal sequence x(n) can be transformed to low frequency and divided into sub-bands. Since the frequency band of the low-pass filter is relatively narrow, the signal can be extracted by a factor M≤K, and the corresponding frequency band spectrum is moved to low frequency and filtered through the low-pass filter to obtain:

[0058]

[0059] Where j represents the complex number representation of the complex exponential.

[0060] The filtered data is sampled M times with a sampling factor of M to obtain:

[0061]

[0062] Therefore, the integrated filter needs to perform the opposite transformation, the input signal The upsampling is performed by a factor of M. After upsampling and interpolation, the output spectrum is compressed M times of the input spectrum, that is, the period of the interpolated spectrum becomes 1 / M of the original. Therefore, on the digital frequency axis, repeated waveforms will be generated within the range of 2π, which is called an image. Then filtering is performed to eliminate the image. Then multiplying by a complex exponential k=0,1,2,...K-1, move the spectrum to high frequency, add up the frequency-transformed outputs of K filters, and you can get the output sequence:

[0063]

[0064] Where g0(n-mM) is the comprehensive time window function and the delay parameter is mM.

[0065] A4. Output the reconstructed language signal through the volume output unit VOLUME.

[0066] The hearing aid system processes the speech signal, and the specific processing steps include: B1, collecting the original speech signal by the speech signal acquisition unit and sending it to the PGA amplification circuit in the analog circuit, amplifying and converting the original speech signal by the PGA amplification circuit and the SD_ADC analog-to-digital conversion circuit in sequence to obtain a digital modulation signal, and sending the digital modulation signal to the multi-channel speech processing device;

[0067] B2. Processing the digital modulated signal using a multi-channel speech processing device and steps A1 to A4 above to obtain a reconstructed speech signal;

[0068] B3. Output the reconstructed language signal through the voice signal output unit.

[0069] The present application converts the original speech signal collected by the microphone into a frequency domain signal by adopting a hearing aid system based on a multi-channel speech processing device, that is, the digital speech signal is divided into multiple frequency band signals of different channels (different frequency bands) through digital sampling in the frequency domain and the first filter group of the multi-channel, and then each frequency band signal is subjected to high-pass filtering, analysis, calculation, noise elimination processing, wide dynamic compression hearing compensation, frequency equalization processing, and synthesis in turn, and finally the synthesized amplified and reconstructed speech signal is output through the speaker, thereby outputting a clear and balanced speech signal. This frequency band audio processing method of amplifying signals of different frequency bands is conducive to improving the effect of speech signal reconstruction, making the reconstructed language speech signal closer to the original speech signal, thereby avoiding the problem of unclear hearing when speaking softly and hearing damage caused by loud sounds. The hearing aid system is precisely adjusted for different hearing-impaired people, meets the hearing compensation needs of different hearing patients, improves the effect of speech signal reconstruction, and improves the wearing and using experience of hearing aids for hearing-impaired people.

[0070] Figure 4 The horizontal axis represents time, and the vertical axis represents the amplitude of the speech signal. Figure 5 The horizontal axis represents the frequency, and the vertical axis represents the amplitude of the speech signal. Figure 4 、 Figure 5 It can be seen that after the voice signal is processed by the hearing aid system of the present application, the reconstructed voice signal obtained is basically consistent with the original voice signal both in the time domain and in the frequency domain. Therefore, the multi-channel voice processing device of the present application can achieve a high degree of restoration of the original voice signal, avoiding the problem of not being able to hear clearly when speaking softly and hearing damage caused by speaking loudly.

[0071] The above are only preferred embodiments of the present application, and the present invention is not limited to the above embodiments. It is understood that other improvements and variations directly derived or imagined by those skilled in the art without departing from the spirit and concept of the invention should be considered to be included in the scope of protection of the invention.

Claims

1. A multi-channel speech processing device, comprising a speech signal input unit, wherein the speech signal input unit inputs a digital modulation signal, wherein: The voice signal input terminal is connected in sequence to the modulator DEM, high-pass filter HPF, analysis filter group WOLA_A, FFT unit, noise elimination unit NR, hearing compensation unit WDRC, frequency equalization unit EQ, IFFT unit, integrated filter group WOLA_S, low-pass filter LPF, and volume output unit VOLUME. The modulator DEM is used to restore the digital modulated signal to a digital voice signal, and the high-pass filter HPF, the analysis filter bank WOLA_A, the FFT unit, the noise elimination unit NR, the hearing compensation unit WDRC, and the frequency equalization unit EQ are used to perform high-pass filtering, analysis, calculation, noise elimination, hearing compensation, and frequency equalization on the digital voice signal, respectively, to obtain a reconstructed speech signal; The volume output unit VOLUME is used to output the reconstructed speech signal; the analysis filter group WOLA_A includes at least one first filter group connected in parallel, which divides the speech signal processed by the high-pass filter HPF into a corresponding number of sub-bands according to the frequency segment definition; the first filter contains K channels: the 0th channel to the K-1th channel, the 0th channel is a low-pass filter, the K-1th channel is a high-pass filter, and the remaining channels are band-pass filters.

2. The multi-channel speech processing device according to claim 1, characterized in that: The integrated filter bank WOLA_S includes at least one second filter bank connected in parallel, and the second filter bank is used to add the speech signals after the FFT unit operation.

3. A method for processing speech signals, the method applying the multi-channel speech processing device according to claim 1, characterized in that: The method comprises: A1, inputting a digital modulation signal through a voice signal input terminal; A2. The digital modulated signal is restored to a digital voice signal through the modulator DEM; A3. The digital speech signal is subjected to high-pass filtering, analysis, fast Fourier transform (FFT) to convert the time domain signal into a frequency domain signal, noise elimination, hearing compensation, frequency equalization, and inverse Fourier transform (IFFT) to convert the frequency domain signal into a time domain signal, respectively, through a high-pass filter (HPF), an analysis filter bank (WOLA_A), an FFT unit, a noise cancellation unit, a hearing compensation unit, a frequency equalization unit, and an IFFT unit, to obtain a reconstructed speech signal. A4. Output the reconstructed speech signal through a volume output unit VOLUME.

4. The speech signal processing method according to claim 3, wherein: In step A3, the input second speech signal is separated by the first filter group in the analysis filter group WOLA_A, and the second speech signal is evenly divided into several frequency band signals. The number of frequency band signals is consistent with the number of channels. After the signal separation, each frequency band signal is sequentially operated, noise eliminated, hearing compensated, and frequency equalized in the corresponding channel.

5. A hearing aid system, encapsulated in a hearing aid housing, the hearing aid system comprising a voice signal acquisition unit, a main control state machine (FSM), an analog circuit, a power module, and a voice signal output unit, wherein the main control state machine (FSM) is used to control the hearing aid system, the analog circuit comprises a PGA amplifier circuit, an SD_ADC analog-to-digital conversion circuit, and an audio power amplifier (PA), the voice signal acquisition unit is sequentially connected to the PGA amplifier circuit and the SD_ADC analog-to-digital conversion circuit, the voice signal acquisition unit is used to acquire original voice signals, the PGA amplifier circuit is used to amplify the original voice signals, the SD_ADC analog-to-digital conversion circuit is used to perform analog-to-digital conversion on the voice signals amplified by the PGA amplifier circuit, and the power module is used to power the hearing aid system, characterized in that: The hearing aid system comprises the multi-channel speech processing device according to claim 1, wherein the input end of the multi-channel speech processing device is connected to the SD_ADC analog-to-digital conversion circuit, and the output end of the multi-channel speech processing device is sequentially connected to the audio power amplifier PA and the speech signal output unit; The audio power amplifier PA is used to amplify the power of the reconstructed speech signal obtained after processing by the multi-channel speech processing device; The speech signal output unit is used to output the amplified and reconstructed speech signal.

6. The hearing aid system according to claim 5, wherein: The master state machine FSM includes an I2C unit, a key detection unit KEYDET, a volume and prompt tone control unit CTRL, and a digital module control unit DMCU. The I2C unit is used to communicate with an off-chip storage unit and directly access internal registers through an off-chip debugging communication interface; the key detection unit KEYDET is connected to a key KEY, and is used to detect a switch signal input by the key KEY; the volume and prompt tone control unit CTRL is used to control the volume of the hearing aid and to superimpose a prompt tone in the reconstructed speech signal; the digital module control unit DMCU is used to convert off-chip debugging control instructions into internal read and write control signals, and to convert the switch signal input by the key KEY into corresponding parameter read and write control signals.

7. The hearing aid system according to claim 6, wherein: The voice signal acquisition unit includes a microphone MIC; the voice signal output unit includes a speaker SPEAK, and the analog circuit also includes a sigma-delta modulator and a power amplifier unit PA. The sigma-delta modulator is used to convert the reconstructed speech signal into a 1-bit modulated signal, and the power amplifier unit PA is used to amplify the 1-bit modulated signal to obtain an amplified voice signal.

8. A speech signal processing method, which uses the multi-channel speech processing device according to claim 1 and the hearing aid system according to claim 5, characterized in that: The method comprises: B1, collecting an original voice signal through a voice signal collection unit and sending the signal to a PGA amplifier circuit in an analog circuit, amplifying and performing analog-to-digital conversion on the original voice signal in sequence through the PGA amplifier circuit and the SD_ADC analog-to-digital conversion circuit to obtain a digital modulation signal, and sending the digital modulation signal to a multi-channel voice processing device; B2. Processing the digital modulated signal through a multi-channel speech processing device to obtain a reconstructed speech signal; B3. Output the reconstructed language signal through the voice signal output unit.

Images