Voice signal processing method and device, electronic equipment, earphone, hearing aid, automobile and medium

Abstract

The invention provides a voice signal processing method and device, electronic equipment, an earphone, a hearing aid, an automobile and a medium. The method comprises the following steps: acquiring a voice observation signal acquired by a voice acquisition device; pre-separating the voice observation signal to obtain a first pre-separated signal and a second pre-separated signal, a first distance between a sound source of the first pre-separated signal and the voice acquisition device being different from a second distance between a sound source of the second pre-separated signal and the voice acquisition device; and performing blind source separation on the voice observation signal according to the first pre-separation signal to obtain a first source voice signal of a sound source of the first pre-separation signal, and performing blind source separation on the voice observation signal according to the second pre-separation signal to obtain a second source voice signal of a sound source of the second pre-separation signal. According to the method and the device, the source voice signals of the sound sources in different distance conditions can be effectively distinguished, so that the sound source separation effect is effectively improved.

Description

Technical Field

[0001] The present disclosure relates to the field of speech processing technology, and in particular to a speech signal processing method, device, electronic device, earphone, hearing aid, automobile and medium. Background Art

[0002] Source separation technology, including separation between different speech signals, separation of speech signals from other signals (music, noise, etc.), is mainly used to solve technical problems caused by the cocktail party effect. The most commonly used source separation technology is the blind source separation algorithm. Blind source separation, also known as blind signal separation, refers to the process of separating each source speech signal from the aliased signal (speech observation signal) when the theoretical model of the signal and the source signal cannot be accurately known. With the continuous development of neural network technology, artificial intelligence technology has been widely used in speech, image, network video, natural language processing, signal processing and other fields.

[0003] In the related art, it is not possible to effectively distinguish the source speech signals of sound sources at different distances, resulting in poor sound source separation effect. Summary of the invention

[0004] The present disclosure aims to solve one of the technical problems in the related art at least to some extent.

[0005] To this end, the purpose of the present disclosure is to propose a speech signal processing method, device, electronic device, earphones, hearing aids, automobiles, non-transitory computer-readable storage media storing computer instructions, and computer program products, which can effectively distinguish source speech signals of sound sources at different distances and effectively improve the sound source separation effect.

[0006] To achieve the above-mentioned purpose, the speech signal processing method proposed in the first aspect embodiment of the present disclosure includes: obtaining a speech observation signal collected by a speech collection device; performing pre-separation processing on the speech observation signal to obtain a first pre-separation signal and a second pre-separation signal, wherein a first distance between a sound source of the first pre-separation signal and the speech collection device and a second distance between a sound source of the second pre-separation signal and the speech collection device are different; and performing blind source separation on the speech observation signal according to the first pre-separation signal to obtain a first source speech signal of the sound source of the first pre-separation signal, and performing blind source separation on the speech observation signal according to the second pre-separation signal to obtain a second source speech signal of the sound source of the second pre-separation signal.

[0007] To achieve the above-mentioned purpose, a speech signal processing device proposed in an embodiment of the second aspect of the present disclosure includes: an acquisition module, used to acquire a speech observation signal collected by a speech collection device; a first separation module, used to perform pre-separation processing on the speech observation signal to obtain a first pre-separation signal and a second pre-separation signal, wherein a first distance between a sound source of the first pre-separation signal and the speech collection device and a second distance between a sound source of the second pre-separation signal and the speech collection device are different; and a second separation module, used to perform blind source separation on the speech observation signal according to the first pre-separation signal to obtain a first source speech signal of the sound source of the first pre-separation signal, and perform blind source separation on the speech observation signal according to the second pre-separation signal to obtain a second source speech signal of the sound source of the second pre-separation signal.

[0008] The electronic device proposed in the third aspect embodiment of the present disclosure includes: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the speech signal processing method proposed in the first aspect embodiment of the present disclosure is implemented.

[0009] The headset proposed in the fourth aspect of the present disclosure includes: a processor; a memory for storing instructions executable by the processor; wherein the processor is configured to: implement the voice signal processing method proposed in the first aspect of the present disclosure.

[0010] The hearing aid proposed in the fifth embodiment of the present disclosure includes: a processor; a memory for storing instructions executable by the processor; wherein the processor is configured to: implement the speech signal processing method proposed in the first embodiment of the present disclosure.

[0011] The automobile proposed in the sixth aspect of the present disclosure includes: a processor; a memory for storing instructions executable by the processor; wherein the processor is configured to: implement the speech signal processing method proposed in the first aspect of the present disclosure.

[0012] The seventh aspect embodiment of the present disclosure proposes a non-temporary computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the speech signal processing method proposed in the first aspect embodiment of the present disclosure is implemented.

[0013] The eighth aspect embodiment of the present disclosure proposes a computer program product. When the instructions in the computer program product are executed by a processor, the speech signal processing method proposed in the first aspect embodiment of the present disclosure is executed.

[0014] The present disclosure provides a speech signal processing method, device, electronic device, earphone, hearing aid, automobile, non-transitory computer-readable storage medium storing computer instructions, and computer program product. The method comprises acquiring a speech observation signal collected by a speech collection device and performing pre-separation processing on the speech observation signal to obtain a first pre-separation signal and a second pre-separation signal, wherein a first distance between a sound source of the first pre-separation signal and the speech collection device and a second distance between a sound source of the second pre-separation signal and the speech collection device are different, and blind source separation is performed on the speech observation signal according to the first pre-separation signal to obtain a first source speech signal of the sound source of the first pre-separation signal, and blind source separation is performed on the speech observation signal according to the second pre-separation signal to obtain a second source speech signal of the sound source of the second pre-separation signal. Since the first pre-separation signal and the second pre-separation signal are obtained based on distance separation, the first distance between the sound source of the first pre-separation signal and the speech collection device and the second distance between the sound source of the second pre-separation signal and the speech collection device are different. Therefore, when blind source separation is subsequently guided based on the first pre-separation signal and the second pre-separation signal, source speech signals of sound sources with different distances can be effectively distinguished, thereby effectively improving the sound source separation effect.

[0015] Additional aspects and advantages of the present disclosure will be given in part in the following description and in part will be obvious from the following description or learned through practice of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The above and / or additional aspects and advantages of the present disclosure will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0017] Figure 1 is a flowchart of a speech signal processing method proposed in an embodiment of the present disclosure;

[0018] Figure 2 is a flow chart of a speech signal processing method proposed in another embodiment of the present disclosure;

[0019] Figure 3 is a schematic diagram of a speech signal processing application in an embodiment of the present disclosure;

[0020] Figure 4 is a flow chart of a speech signal processing method proposed in another embodiment of the present disclosure;

[0021] Figure 5 is a flow chart of a speech signal processing method proposed in another embodiment of the present disclosure;

[0022] Figure 6 is a structural diagram of a speech signal processing device proposed in an embodiment of the present disclosure;

[0023] Figure 7 A block diagram of an exemplary electronic device suitable for implementing the embodiments of the present disclosure is shown;

[0024] Figure 8 It is a schematic diagram of a functional block diagram of a car shown in an exemplary embodiment. DETAILED DESCRIPTION

[0025] Embodiments of the present disclosure are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present disclosure, and are not to be construed as limitations of the present disclosure. On the contrary, the embodiments of the present disclosure include all changes, modifications, and equivalents that fall within the spirit and connotation of the appended claims.

[0026] Figure 1 It is a flowchart of a speech signal processing method proposed in an embodiment of the present disclosure.

[0027] This embodiment takes the voice signal processing method configured as a voice signal processing device as an example. In this embodiment, the voice signal processing method can be configured in the voice signal processing device, and the voice signal processing device can be set in a server, or can also be set in an electronic device. The embodiment of the present disclosure does not limit this.

[0028] The speech signal processing method in this embodiment can be configured in an electronic device, for example, wherein the electronic device is a hardware device with various operating systems, such as a smart phone, a tablet computer, a personal digital assistant, an e-book, a vehicle-mounted device, or the electronic device can also be a hearing aid, a headset, a telephone device, a sound reinforcement system, etc., without limitation.

[0029] It should be noted that the execution subject of the embodiment of the present disclosure may be, for example, a central processing unit (CPU) in a server or electronic device in terms of hardware, or may be, for example, a related background service in a server or electronic device in terms of software, without limitation.

[0030] like Figure 1 As shown, the speech signal processing method comprises:

[0031] S101: Acquire a speech observation signal collected by a speech collection device.

[0032] Among them, the device that can be used to collect and pick up voice signals can be called a voice collection device. The voice collection device can be, for example, a voice collection unit, such as a microphone, or the voice collection device can also be, for example, a voice collection array. The voice collection array is composed of a number of orderly arranged voice collection units. The voice collection unit can be, for example, a microphone, and there is no limitation to this.

[0033] In some embodiments, the speech observation signal may refer to a sound signal collected by a speech collector, and there is no limitation to this.

[0034] In some embodiments, the speech observation signal may be a speech observation signal collected by a speech collection device alone. If the speech collection device includes multiple speech collection units, the speech observation signal may also include speech observation signals collected by each speech collection unit respectively, and there is no limitation on this.

[0035] S102: Pre-separating the speech observation signal to obtain a first pre-separated signal and a second pre-separated signal, wherein a first distance between a sound source of the first pre-separated signal and a speech collection device is different from a second distance between a sound source of the second pre-separated signal and the speech collection device.

[0036] In some embodiments, after obtaining the voice observation signal collected by the voice collection device, the voice observation signal can be first pre-separated. The different voice signals obtained by the pre-separation can be respectively referred to as the first pre-separation signal and the second pre-separation signal, without limitation.

[0037] In some embodiments, the speech observation signal may be pre-separated based on the distance to identify from the speech observation signal a speech signal obtained by collecting sound signals of source speech signals of sound sources at different distances, without limitation.

[0038] In some embodiments, the distance condition refers to the distance between the sound source and the voice collection device, and there is no limitation on this.

[0039] In some embodiments, the speech observation signal is pre-separated to obtain a first pre-separated signal and a second pre-separated signal. The first distance between the sound source of the first pre-separated signal and the speech collection device and the second distance between the sound source of the second pre-separated signal and the speech collection device are different, and there is no limitation on this.

[0040] In some embodiments, the first distance between the sound source of the first pre-separated signal and the voice collection device may be a relatively close distance, and the second distance between the sound source of the second pre-separated signal and the voice collection device may be a relatively long distance, or the first distance between the sound source of the first pre-separated signal and the voice collection device may be a relatively long distance, and the second distance between the sound source of the second pre-separated signal and the voice collection device may be a relatively close distance, without limitation.

[0041] In some embodiments, before performing blind source separation, the speech observation signal can be pre-separated based on a pre-separation module to obtain a first pre-separation signal and a second pre-separation signal. Since the first pre-separation signal and the second pre-separation signal are obtained based on distance separation, the first distance between the sound source of the first pre-separation signal and the speech collection device and the second distance between the sound source of the second pre-separation signal and the speech collection device are different. Therefore, when the blind source separation is subsequently guided by the first pre-separation signal and the second pre-separation signal, the source speech signals of sound sources at different distances can be effectively distinguished, thereby effectively improving the sound source separation effect.

[0042] S103: Perform blind source separation on the speech observation signal according to the first pre-separation signal to obtain a first source speech signal of the sound source of the first pre-separation signal, and perform blind source separation on the speech observation signal according to the second pre-separation signal to obtain a second source speech signal of the sound source of the second pre-separation signal.

[0043] The sound source of the first pre-separated signal and the sound source of the second pre-separated signal are different sound sources, and the distances between the different sound sources and the voice collection device are different.

[0044] In some embodiments, after pre-separation processing is performed on the speech observation signal to obtain a first pre-separation signal and a second pre-separation signal, the first pre-separation signal can be used to guide the process of blind source separation of the speech observation signal to obtain a first source speech signal of the sound source of the first pre-separation signal, without any limitation.

[0045] In some embodiments, after pre-separation processing is performed on the speech observation signal to obtain a first pre-separation signal and a second pre-separation signal, the second pre-separation signal can also be used to guide the process of blind source separation of the speech observation signal to obtain a second source speech signal of the sound source of the second pre-separation signal. There is no limitation on this.

[0046] Therefore, in this embodiment, by acquiring the voice observation signal collected by the voice collection device and performing pre-separation processing on the voice observation signal, a first pre-separation signal and a second pre-separation signal are obtained, wherein the first distance between the sound source of the first pre-separation signal and the voice collection device and the second distance between the sound source of the second pre-separation signal and the voice collection device are different, and the voice observation signal is blindly source separated according to the first pre-separation signal to obtain the first source voice signal of the sound source of the first pre-separation signal, and the voice observation signal is blindly source separated according to the second pre-separation signal to obtain the second source voice signal of the sound source of the second pre-separation signal. Since the first pre-separation signal and the second pre-separation signal are obtained based on distance separation, the first distance between the sound source of the first pre-separation signal and the voice collection device and the second distance between the sound source of the second pre-separation signal and the voice collection device are different, so that when blind source separation is guided based on the first pre-separation signal and the second pre-separation signal, the source voice signals of the sound sources in different distance situations can be effectively distinguished, thereby effectively improving the sound source separation effect.

[0047] The application scenarios of the speech signal processing method in the embodiment of the present disclosure are illustrated as follows:

[0048] Application scenario A: It can be used in scenarios such as hearing aids, headphones, telephones, online conferences, sound reinforcement systems, and vehicle-mounted equipment, which can effectively improve the voice function experience of related products in the application scenario.

[0049] For example, the user may only want to hear the voice of the user in front of him who is talking to him. Therefore, by adopting the voice processing method in the embodiment of the present disclosure, the hearing aid function of the hearing aid can effectively distinguish the voices of users at different distances.

[0050] Application scenario B: Noise reduction for calls using headphones, phones, or online conferences.

[0051] For example, when making a call, users hope that local devices such as headphones and phones will only pick up their own voices; as more and more users communicate through online conferences at home, they also hope that the conference terminal (usually a laptop) will only pick up their own voices.

[0052] Application scenario C: acoustic feedback suppression in sound reinforcement systems.

[0053] For example, sound reinforcement systems are widely used, and by extracting close-range sounds, the howling problem of the sound reinforcement system can be effectively solved.

[0054] Application scenario D: external noise suppression.

[0055] For example, when the car windows are open, whether it is an in-car call or in-car voice interaction, it is hoped that the microphone in the car can block out the noise or human voice outside the car. Through distance-based sound source separation, the sound outside the car can be identified as long-distance noise and blocked.

[0056] In some embodiments of the present disclosure, the speech collection device may include multiple speech collection units, and the speech observation signal includes at least: a first speech observation signal collected by a first speech collection unit, and the first speech collection unit belongs to multiple speech collection units. The speech observation signal is pre-separated to obtain a first pre-separated signal and a second pre-separated signal. The first speech observation signal is pre-separated to obtain a first pre-separated signal and a second pre-separated signal, thereby improving the efficiency and rationality of the pre-separation.

[0057] In some embodiments of the present disclosure, randomly selecting a voice collection unit from a plurality of voice collection units as the first voice collection unit can effectively expand the application scenario and effectively avoid excessive resource consumption by pre-separation.

[0058] Figure 2 It is a flowchart of a speech signal processing method proposed in another embodiment of the present disclosure.

[0059] like Figure 2 As shown, the speech signal processing method comprises:

[0060] S201: Acquire a speech observation signal collected by a speech collection device, where the speech collection device includes multiple speech collection units, and the speech observation signal includes at least: a first speech observation signal collected by a first speech collection unit, where the first speech collection unit belongs to the multiple speech collection units.

[0061] S202: Input the first speech observation signal into the pre-separation model to obtain the first pre-separation signal and the second pre-separation signal output by the pre-separation model, wherein the pre-separation model is obtained by deep learning training using a training set, the training set comes from multiple speech acquisition units, the training set includes multiple samples, one speech acquisition unit corresponds to at least one sample, each sample includes: a sample observation signal collected by the speech acquisition unit, and a first sample speech signal and a second sample speech signal corresponding to the sample observation signal, the third distance between the sound source of the first sample speech signal and the speech acquisition unit and the fourth distance between the sound source of the second sample speech signal and the speech acquisition unit are different.

[0062] That is to say, the initial neural network model can be pre-trained based on the deep learning method. The initial neural network model has the function of pre-separating the speech observation signal based on distance. During the training process, if it is determined that the neural network model converges during training, the converged neural network model can be used as the pre-separation model. The converged neural network model can be a neural network model obtained by iteratively training the initial neural network model. There is no restriction on this.

[0063] Among them, the structure of the initial neural network model can flexibly adopt convolutional neural network (CNN), long short-term memory network (LSTM), etc., without any restriction.

[0064] In some embodiments, a training set can be first obtained, where the training set includes multiple samples, each sample including: a sample observation signal collected by a speech collection unit, and a first sample speech signal and a second sample speech signal corresponding to the sample observation signal, the third distance between the sound source of the first sample speech signal and the speech collection unit and the fourth distance between the sound source of the second sample speech signal and the speech collection unit are different, and there is no limitation on this.

[0065] The first sample speech signal and the second sample speech signal refer to speech signals at different distances in the sample observation signal, and there is no limitation on this.

[0066] In some embodiments, the first sample speech signal and the second sample speech signal can be obtained by pre-marking the sample observation signals, and the third distance between the sound source of the first sample speech signal and the speech collection device and the fourth distance between the sound source of the second sample speech signal and the speech collection device are different, and there is no limitation on this.

[0067] In some embodiments, the third distance between the sound source of the first sample voice signal and the voice collection device may be a relatively close distance, and the fourth distance between the sound source of the second sample voice signal and the voice collection device may be a relatively long distance, or the third distance between the sound source of the first sample voice signal and the voice collection device may be a relatively long distance, and the fourth distance between the sound source of the second sample voice signal and the voice collection device may be a relatively close distance, without limitation.

[0068] In some embodiments, the process of obtaining a pre-separation model through deep learning training can be referred to in related technologies and will not be described in detail here.

[0069] S203: Perform blind source separation on the speech observation signal according to the first pre-separation signal to obtain a first source speech signal of the sound source of the first pre-separation signal, and perform blind source separation on the speech observation signal according to the second pre-separation signal to obtain a second source speech signal of the sound source of the second pre-separation signal.

[0070] like Figure 3 As shown, Figure 3 is a schematic diagram of the speech signal processing application in the embodiment of the present disclosure. In the first stage (pre-separation stage), the speech observation signal can be pre-separated. The speech acquisition device can include two speech acquisition units, each of which is a microphone, for example. The speech observation signal x1 collected by the first microphone can be pre-separated. The pre-separation process can be implemented based on a pre-separation model. For example, the speech observation signal x1 is input into the pre-separation model to obtain a first pre-separation signal Second pre-separation signal Then, based on the first pre-separation signal and the second pre-separation signal Instruct the Independent Vector Analysis (IVA) algorithm to perform blind source separation on the speech observation signals x1 and x2 of the speech acquisition device.

[0071] In some embodiments, if the speech collection device includes two speech collection units, and the two speech collection units collect speech observation signals x1 and speech observation signals x2 respectively, then based on the first pre-separation signal Instruct one of the speech observation signals (such as speech observation signal x1, or speech observation signal x2) to separate the first pre-separation signal The source speech signal of the sound source (which may be referred to as the first source speech signal) may be obtained based on the second pre-separation signal. Instruct one of the speech observation signals (such as speech observation signal x1, or speech observation signal x2) to separate the second pre-separation signal The source speech signal of the sound source (which can be called the second source speech signal) is not limited to this.

[0072] In the first stage pre-separation process, the goal is to obtain a close-range signal (an optional example of the first pre-separation signal) and a long-range signal (an optional example of the second pre-separation signal). The input of the pre-separation module is the mixed signal to be separated received by a single microphone (an optional example of the first voice acquisition unit), and the output is the close-range signal and the long-range signal after preliminary separation. The pre-separation module can be a network model trained using a deep learning method.

[0073] The following is an example of the model training method:

[0074]

[0075]

[0076] Where f is the separation network, X mix , X near , X far They are the mixed signal to be separated, the clean close-range signal and the clean long-range signal. They are the separated close-range and long-range signals, respectively. The aforementioned signals can be time domain signals or frequency domain signals after short-time Fourier transform. Loss represents the loss function used in training. Any form of loss function can be flexibly used without restriction. The training set can be generated by simulation or actual recording. The training set mainly consists of multiple close-range signals X in the room. near and multiple long-distance signals X far , and the mixture of the two forms X mix composition.

[0077] It should be noted that blind source separation (BSS) is a technique that, based on the assumption that source signals are independent of each other, can recover the source signal according to certain criteria using the observation signal obtained after mixing when the source signal and the mixed system are unknown. The speech observation signal collected by the speech collection device in the disclosed embodiment is a mixed signal of multiple sound signals, and the blind source separation technique can separate the multiple sound signals according to the speech observation signal.

[0078] In the disclosed embodiment, the first pre-separation signal and the second pre-separation signal obtained based on distance separation can be used to guide the blind source separation process, and the prior pilot information can be used to fix the close-range signal (an optional example of the first pre-separation signal) and the long-range signal (an optional example of the second pre-separation signal) to the corresponding channels, which can not only solve the technical problems of leakage and distortion of the sound source separation technology based on the neural network, but also effectively solve the technical problem that the technology combining beamforming and blind source separation cannot separate the source speech signals of the sound sources in the same orientation.

[0079] Figure 4 It is a flowchart of a speech signal processing method proposed in another embodiment of the present disclosure.

[0080] like Figure 4 As shown, the speech signal processing method comprises:

[0081] S401: Acquire a speech observation signal collected by a speech collection device.

[0082] The speech observation signal may be a speech observation signal collected by any one of the multiple speech collection units. Figure 3 The speech observation signal x1 is shown in , and there is no limitation on this.

[0083] S402: Pre-separate the speech observation signal to obtain a first pre-separation signal and a second pre-separation signal, wherein a first distance between a sound source of the first pre-separation signal and a speech collection device is different from a second distance between a sound source of the second pre-separation signal and the speech collection device.

[0084] The pre-separated speech observation signal may be the above Figure 3 The speech observation signal x1 or the speech observation signal x2 shown in the figure, the first pre-separation signal obtained by separation can be used to guide the blind source separation of the speech observation signal x1, and there is no limitation on this.

[0085] This embodiment provides an implementation method for blind source separation of a speech observation signal according to a first pre-separation signal to obtain a first source speech signal of the sound source of the first pre-separation signal. That is, the first pre-separation signal is fixed to a corresponding channel (for example, a channel for outputting the speech observation signal x1, or a channel for outputting the speech observation signal x2) to achieve effective separation of the source speech signal of the sound source at a distance corresponding to the first pre-separation signal, thereby effectively improving the sound source separation effect.

[0086] S403: Determine a variance term of a probability density function of a sound source corresponding to the speech observation signal.

[0087] Assume that there are two sound sources, one inside and one outside the specified distance, with no restrictions on the direction.

[0088] In the embodiment of the present disclosure, a dual-microphone array (without limitation on the form of the voice collection device) may be used to receive the voice observation signal, and after short-time Fourier transform, the frequency domain representation of the microphone signal (an optional example of the voice observation signal), the original sound source signal, and the estimated sound source signal (an optional example of the source voice signal) may be obtained:

[0089] X(t,f)=[X1(t,f),X2(t,f)] T ;

[0090] S(t,f)=[S1(t,f),S2(t,f)] T ;

[0091] Y(t,f)=[Y1(t,f),Y2(t,f)] T ;

[0092] Where t represents time, f represents frequency, and the superscript T represents vector transpose.

[0093] The original sound source signal is received by the microphone after propagation. For example, it can be mixed and separated in the frequency domain by a convolution mixing system. The frequency domain mixing model and separation model are expressed as follows:

[0094] X(ω,n)=A(f)S(t,f);

[0095] Y(ω,n)=W(f)X(t,f);

[0096] Where A(f) and W(f) are the mixing matrix and separation matrix respectively. The blind source separation algorithm estimates the separation matrix W(f) without any prior information, and uses the mixed signal X(t,f) (an optional example of a speech observation signal) to estimate the sound source signal Y(t,f) (an optional example of a source speech signal).

[0097] In the blind source separation process, it is assumed that each sound source is independent of each other. The IVA algorithm assumes a multivariate probability density function to utilize the correlation between frequencies, which not only retains the internal dependence between the frequencies of each source signal, but also maximizes the independence between different sound source signals to effectively avoid the arrangement problem, thus ensuring that the separated signal is consistent across the entire frequency band. Based on this, the objective function of the IVA algorithm is:

[0098]

[0099] Among them, p(Y k (t,f)) is the multivariate probability density function of the sound source, 1≤k≤T, k=1,2, T is the number of sound signal (speech observation signal) frames in the frequency domain, E[] represents the mathematical expectation, log represents the logarithm, detW(f) represents the determinant of the matrix W(f), represents the modulus, f represents the frequency point mark, and F represents the last frequency point mark.

[0100] Assuming that the sound source obeys a time-varying Gaussian distribution, the probability density function of the sound source p(Y k (t,f)) satisfies:

[0101]

[0102] Among them, Y k (t) represents the k (t,f) is normalized on the frequency marker f, σ k 2 (t) is the time-varying variance (variance term) of the probability density function of the sound source, || ||2 represents the L2 norm, The representation is to square the L2 norm, ∝ represents a label that obeys a certain distribution, 1≤k≤T, k=1,2, T is the number of acoustic signal (speech observation signal) frames in the frequency domain. Based on this probability density function, IVA solves the frequency permutation problem of blind source separation.

[0103] In some embodiments, the variance term of the probability density function of the sound source corresponding to the speech observation signal is determined, which may be σ k 2 (t) without limitation.

[0104] S404: Using the first pre-separated signal as a pilot signal of a variance term of a probability density function of a sound source, so as to obtain a variance term of a probability density function of a sound source into which the pilot signal is introduced.

[0105] In the embodiment of the present disclosure, in order to solve the technical problem of global replacement, the pre-separation signal obtained in the first stage can be used, which is denoted as Y p (t,f)=[Y p,1 (t,f),Y p,2 (t,f)] T , Y p,1 (t,f) and Y p,2 (t, f) are short-distance signals and long-distance signals divided by specific distances, respectively. Their energy (signal) is added to the time-varying variance of the probability density function of the sound source to obtain a new probability density function:

[0106]

[0107] Wherein, γ is the weight coefficient of the pre-separation signal (the first pre-separation signal or the second pre-separation signal), σ p,k 2 When p in (t) is 1, it means the short-distance signal Y is divided by a specific distance. p,1 The energy of (t,f), σ p,k 2 When p is 2 in (t), it means the short-distance signal Y is divided by a specific distance. p,2 (t, f) energy. In this embodiment, since the first pre-separated signal is fixed to the corresponding channel, σ p,k 2 (t) where p is 1, for σ p,k 2 The implementation method in which p is 2 in (t) can be found in the following examples.

[0108] In some embodiments of the present disclosure, the variance term of the probability density function of the sound source of the pilot signal may be, for example, γ 2 σ p,k 2 (t) without limitation.

[0109] S405: Perform blind source separation on the speech observation signal according to the first separation matrix to obtain an initial separation signal frequency vector.

[0110] The first separation matrix may be an Nth generation separation matrix in the IVA algorithm, where N is an integer greater than zero. When N is 1, the first separation matrix is ​​a random matrix, and there is no restriction on this.

[0111] In some embodiments, the speech observation signal can be blindly separated based on the first separation matrix, and the obtained separation signal frequency vector (the meaning of the separation signal frequency vector can be found in the IVA algorithm and will not be repeated here) can be called the initial separation signal frequency vector.

[0112] S406: Determine a first separation signal frequency vector according to the initial separation signal frequency vector, the probability density function variance term of the sound source introducing the pilot signal, and the first separation matrix.

[0113] The separation signal frequency vector used to determine the first source speech signal may be referred to as a first separation signal frequency vector.

[0114] In some embodiments, whether to use the initial separation signal frequency vector as the first separation signal frequency vector may be determined based on the objective function of the IVA algorithm, and there is no limitation to this.

[0115] It can be understood that based on the principle of blind source separation algorithm, according to the assumption that the first separation signal frequency vector (corresponding to the first source voice signal) and the second separation signal frequency vector (corresponding to the second source voice signal) are independent of each other, the separation signal frequency vector recovery is achieved by making the separation signal frequency vectors of different source voice signals as independent as possible. It includes two steps: first, determining the objective function, and using the objective function as a criterion for judging whether the separation signal frequency vector is close to statistical independence, and secondly determining the optimization algorithm, which is used to update the next separation matrix according to the previous separation matrix so that the separation signal frequency vector is close to the statistical independence standard.

[0116] In some embodiments of the present disclosure, if the initial separation signal frequency vector meets a preset condition, the initial separation signal frequency vector is used as the first separation signal frequency vector; if the initial separation signal frequency vector does not meet the preset condition, the reference item is updated according to the first separation matrix, and the updated reference item is obtained, wherein the reference item includes: a probability density function variance item of the sound source of the introduced pilot signal, the first separation matrix is ​​related to the initial reference item, the second separation matrix is ​​determined according to the updated reference item, and the speech observation signal is blind source separated according to the second separation matrix until the separation signal frequency vector obtained by the blind source separation meets the preset condition, and the obtained separation signal frequency vector is used as the first separation signal frequency vector, so as to accurately determine the first separation signal frequency vector to assist in obtaining an accurate first source speech signal.

[0117] In some embodiments, the first separation matrix can be determined based on a reference item, and the second separation matrix can be determined based on an updated reference item. If the first separation matrix refers to an Nth generation separation matrix, the second separation matrix can refer to an N+1th generation separation matrix, and there is no limitation on this.

[0118] The expressions and update methods of the reference items are as follows:

[0119] (1) Update the weighted covariance matrix V k (f):

[0120]

[0121] Among them, r k (t) represents the L2 norm of the first source speech signal.

[0122]

[0123] in, Represents the variance estimate of the kth sound source.

[0124]

[0125] Among them, φ(r k (t)) represents the coefficient of the weighted covariance matrix after the pilot signal is introduced, φ(r k (t)) can be an optional example of a reference item, which is not limited to this. The reference item includes: the variance term γ of the probability density function of the sound source of the pilot signal 2 σ p,k 2 (t).

[0126]

[0127] Among them, V k(f) represents the weighted covariance matrix, which can be used to calculate new reference items according to the first separation matrix based on the above formula, and use the new reference items as updated reference items, without limitation.

[0128] (2) Update the separation matrix ω k (f):

[0129] ω k (f)←(W(f)V k (f)) -1 e k ;

[0130] Among them, ω k (f) represents a matrix element in the separation matrix, that is, for each ω k (f) Update.

[0131]

[0132] Among them, e k is a 4×1 unit vector, the nth element is 1, the other elements are 0, H represents the conjugate transpose, () -1 Denotes matrix inversion, and the second separation matrix can be determined according to the updated reference item based on the formula shown in the above (2), and there is no limitation on this.

[0133] After updating the reference item according to the first separation matrix, obtaining the updated reference item, and determining the second separation matrix according to the updated reference item, the speech observation signal can be blind source separated according to the second separation matrix until the separation signal frequency vector obtained by the blind source separation meets the preset conditions (for example, the statistical independence standard), and the obtained separation signal frequency vector is used as the first separation signal frequency vector.

[0134] S407: Determine a first source speech signal according to the first separated signal frequency vector.

[0135] After the first separated signal frequency vector that meets the statistical independence criterion is determined, the first source speech signal can be synthesized according to the first separated signal frequency vector, and there is no limitation to this.

[0136] In this embodiment, the first pre-separation signal is fixed to the corresponding channel, so that the source speech signal of the sound source at the corresponding distance to the first pre-separation signal can be effectively separated, thereby effectively improving the sound source separation effect.

[0137] Figure 5 It is a flowchart of a speech signal processing method proposed in another embodiment of the present disclosure.

[0138] like Figure 5 As shown, the speech signal processing method comprises:

[0139] S501: Acquire a speech observation signal collected by a speech collection device.

[0140] The speech observation signal may be a speech observation signal collected by any one of the multiple speech collection units. Figure 3 The speech observation signal x2 is shown in , and there is no limitation on this.

[0141] S502: Pre-separate the speech observation signal to obtain a first pre-separation signal and a second pre-separation signal, wherein a first distance between a sound source of the first pre-separation signal and a speech collection device is different from a second distance between a sound source of the second pre-separation signal and the speech collection device.

[0142] The pre-separated speech observation signal may be the above Figure 3 The speech observation signal x1 or the speech observation signal x2 shown in the figure, the first pre-separation signal obtained by separation can be used to guide the blind source separation of the speech observation signal x2, and there is no limitation on this.

[0143] This embodiment provides an implementation method for blind source separation of a speech observation signal according to a second pre-separation signal to obtain a second source speech signal of the sound source of the second pre-separation signal. That is to say, the second pre-separation signal is fixed to a corresponding channel (for example, a channel for outputting the speech observation signal x2, or a channel for outputting the speech observation signal x1, and the channels used to separate the first source speech signal and the second source speech signal are different), so as to achieve effective separation of the source speech signal of the sound source at a distance corresponding to the second pre-separation signal, thereby effectively improving the sound source separation effect.

[0144] It should be noted that, in the present embodiment, the second pre-separation signal is fixed to the corresponding channel (for example, the channel for outputting the speech observation signal x2, or the channel for outputting the speech observation signal x1, and the channels used to separate the first source speech signal and the second source speech signal are different). The speech signal processing method adopted, the method for introducing the pilot signal, the first separation matrix, the second separation matrix, the expression of the reference item, and the terminology explanation or implementation method for determining the second separation matrix based on the updated reference item can all be referred to the description in the above embodiments and will not be repeated here.

[0145] S503: Determine the variance term of the probability density function of the sound source corresponding to the speech observation signal.

[0146] S504: Using the second pre-separated signal as a pilot signal of a variance term of a probability density function of a sound source, so as to obtain a variance term of a probability density function of a sound source into which the pilot signal is introduced.

[0147] S505: Perform blind source separation on the speech observation signal according to the first separation matrix to obtain an initial separation signal frequency vector.

[0148] S506: Determine a second separation signal frequency vector according to the initial separation signal frequency vector, the probability density function variance term of the sound source introducing the pilot signal, and the first separation matrix.

[0149] The separation signal frequency vector used to determine the second source speech signal may be referred to as a second separation signal frequency vector.

[0150] In some embodiments of the present disclosure, if the initial separation signal frequency vector meets a preset condition, the initial separation signal frequency vector is used as the second separation signal frequency vector; if the initial separation signal frequency vector does not meet the preset condition, the reference item is updated according to the first separation matrix, and the updated reference item is obtained, wherein the reference item includes: a probability density function variance item of the sound source of the introduced pilot signal, the first separation matrix is ​​related to the initial reference item, the second separation matrix is ​​determined according to the updated reference item, and the speech observation signal is blind source separated according to the second separation matrix until the separation signal frequency vector obtained by the blind source separation meets the preset condition, and the obtained separation signal frequency vector is used as the second separation signal frequency vector, so as to accurately determine the second separation signal frequency vector to assist in obtaining an accurate second source speech signal.

[0151] S507: Determine a second source speech signal according to the second separated signal frequency vector.

[0152] After the second separated signal frequency vector that meets the statistical independence criterion is determined, the first source speech signal can be synthesized according to the second separated signal frequency vector, and there is no limitation to this.

[0153] In this embodiment, the source speech signal of the sound source at the corresponding distance to the second pre-separation signal is effectively separated, thereby effectively improving the sound source separation effect.

[0154] Figure 6 It is a structural diagram of a speech signal processing device proposed in an embodiment of the present disclosure.

[0155] like Figure 6 As shown, the speech signal processing device 60 includes:

[0156] The acquisition module 601 is used to acquire the speech observation signal collected by the speech collection device.

[0157] The first separation module 602 is used to perform pre-separation processing on the speech observation signal to obtain a first pre-separation signal and a second pre-separation signal, wherein a first distance between the sound source of the first pre-separation signal and the speech collection device is different from a second distance between the sound source of the second pre-separation signal and the speech collection device.

[0158] The second separation module 603 is used to perform blind source separation on the speech observation signal according to the first pre-separation signal to obtain a first source speech signal of the sound source of the first pre-separation signal, and to perform blind source separation on the speech observation signal according to the second pre-separation signal to obtain a second source speech signal of the sound source of the second pre-separation signal.

[0159] It should be noted that the above explanation of the speech signal processing method is also applicable to the speech signal processing device of this embodiment and will not be repeated here.

[0160] In this embodiment, a first pre-separation signal and a second pre-separation signal are obtained by acquiring a speech observation signal collected by a speech collection device and performing pre-separation processing on the speech observation signal, wherein a first distance between a sound source of the first pre-separation signal and the speech collection device and a second distance between a sound source of the second pre-separation signal and the speech collection device are different, and blind source separation is performed on the speech observation signal according to the first pre-separation signal to obtain a first source speech signal of the sound source of the first pre-separation signal, and blind source separation is performed on the speech observation signal according to the second pre-separation signal to obtain a second source speech signal of the sound source of the second pre-separation signal. Since the first pre-separation signal and the second pre-separation signal are obtained based on distance separation, the first distance between the sound source of the first pre-separation signal and the speech collection device and the second distance between the sound source of the second pre-separation signal and the speech collection device are different, so that when blind source separation is guided based on the first pre-separation signal and the second pre-separation signal, the source speech signals of sound sources with different distances can be effectively distinguished, thereby effectively improving the sound source separation effect.

[0161] Figure 7 A block diagram of an exemplary electronic device suitable for implementing embodiments of the present disclosure is shown. Figure 7 The electronic device 12 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0162] like Figure 7 As shown, the electronic device 12 is in the form of a general purpose computing device. The components of the electronic device 12 may include, but are not limited to: one or more processors or processing units 16, a memory 28, and a bus 18 that connects various system components (including the memory 28 and the processing unit 16).

[0163] The bus 18 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration port, a processor or a local bus using any of a variety of bus structures. For example, these architectures include but are not limited to Industry Standard Architecture (ISA) bus, Micro Channel Architecture (MAC) bus, Enhanced ISA bus, Video Electronics Standards Association (VESA) local bus and Peripheral Component Interconnection (PCI) bus.

[0164] The electronic device 12 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the electronic device 12, including volatile and non-volatile media, removable and non-removable media.

[0165] The memory 28 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache 32. The electronic device 12 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the storage system 34 may be used to read and write non-removable, non-volatile magnetic media ( Figure 7 Not shown, often called a "hard drive").

[0166] although Figure 7 Not shown in the figure, a disk drive for reading and writing a removable non-volatile disk (e.g., a "floppy disk"), and an optical disk drive for reading and writing a removable non-volatile optical disk (e.g., a compact disc read only memory (hereinafter referred to as: CD-ROM), a digital versatile disc read only memory (hereinafter referred to as: DVD-ROM) or other optical media) may be provided. In these cases, each drive may be connected to the bus 18 via one or more data medium interfaces. The memory 28 may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of the various embodiments of the present disclosure.

[0167] A program / utility 40 having a set (at least one) of program modules 42 may be stored, for example, in the memory 28, such program modules 42 including but not limited to an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. The program modules 42 generally perform the functions and / or methods of the embodiments described in the present disclosure.

[0168] The electronic device 12 may also communicate with one or more external devices 14 (e.g., keyboard, pointing device, display 24, etc.), may also communicate with one or more devices that enable the human body to interact with the electronic device 12, and / or communicate with any device that enables the electronic device 12 to communicate with one or more other computing devices (e.g., network card, modem, etc.). Such communication may be performed through an input / output (I / O) interface 22. In addition, the electronic device 12 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) through a network adapter 20. As shown, the network adapter 20 communicates with other modules of the electronic device 12 through a bus 18. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 12, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0169] The processing unit 16 executes various functional applications and data processing by running the programs stored in the memory 28, such as implementing the speech signal processing method mentioned in the above embodiment.

[0170] An embodiment of the present disclosure further provides a headset, comprising: a processor; and a memory for storing instructions executable by the processor; wherein the processor is configured to: implement the steps of the method in the above embodiment.

[0171] The embodiments of the present disclosure further provide a hearing aid, comprising: a processor; and a memory for storing instructions executable by the processor; wherein the processor is configured to: implement the steps of the method in the above embodiments.

[0172] Figure 8 It is a schematic diagram of a functional block diagram of a car shown in an exemplary embodiment.

[0173] For example, the car 800 may be a hybrid car, a non-hybrid car, an electric car, a fuel cell car, or other types of cars. The car 800 may be an autonomous car, a semi-autonomous car, or a non-autonomous car.

[0174] Reference Figure 8 , the automobile 800 may include various subsystems, for example, an infotainment system 810, a perception system 820, a decision control system 830, a drive system 840, and a computing platform 850. The automobile 800 may also include more or fewer subsystems, and each subsystem may include multiple components. In addition, each subsystem and each component of the automobile 800 may be interconnected by wire or wireless means.

[0175] In some embodiments, the infotainment system 810 may include a communication system, an entertainment system, and a navigation system, etc.

[0176] The perception system 820 may include several sensors for sensing information about the environment around the car 800. For example, the perception system 820 may include a global positioning system (the global positioning system may be a GPS system, or a Beidou system or other positioning systems), an inertial measurement unit (IMU), a laser radar, a millimeter wave radar, an ultrasonic radar, and a camera.

[0177] The decision control system 830 may include a computing system, a vehicle controller, a steering system, a throttle, and a braking system.

[0178] The drive system 840 may include components that provide powered motion for the automobile 800. In one embodiment, the drive system 840 may include an engine, an energy source, a transmission system, and wheels. The engine may be one or a combination of an internal combustion engine, an electric motor, and an air compression engine. The engine is capable of converting energy provided by the energy source into mechanical energy.

[0179] Some or all functions of the automobile 800 are controlled by a computing platform 850. The computing platform 850 may include at least one processor 851 and a memory 852, and the processor 851 may execute instructions 853 stored in the memory 852.

[0180] The processor 851 may be any conventional processor, such as a commercially available CPU. The processor may also include a graphics processor (Graphic Process Unit, GPU), a field programmable gate array (Field Programmable Gate Array, FPGA), a system on chip (System on Chip, SOC), an application specific integrated circuit (Application Specific Integrated Circuit, ASIC) or a combination thereof.

[0181] The memory 852 may be implemented by any type of volatile or nonvolatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0182] In addition to instructions 853 , memory 852 may also store data, such as road maps, route information, and data on the location, direction, speed, etc. of the vehicle. The data stored in memory 852 may be used by computing platform 850 .

[0183] In the embodiment of the present disclosure, the processor 851 may execute instruction 853 to complete all or part of the steps of the above-mentioned speech signal processing method.

[0184] In order to implement the above embodiments, the present disclosure further proposes a non-temporary computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the speech signal processing method proposed in the above embodiments of the present disclosure is implemented.

[0185] In order to implement the above embodiments, the present disclosure further proposes a computer program product. When the instructions in the computer program product are executed by a processor, the speech signal processing method proposed in the above embodiments of the present disclosure is executed.

[0186] It should be noted that, in the description of the present disclosure, the terms "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance. In addition, in the description of the present disclosure, unless otherwise specified, the meaning of "plurality" is two or more.

[0187] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code that includes one or more executable instructions for implementing the steps of a specific logical function or process, and the scope of the preferred embodiments of the present disclosure includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present disclosure belong.

[0188] It should be understood that the various parts of the present disclosure can be implemented in hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0189] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0190] In addition, each functional unit in each embodiment of the present disclosure may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0191] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc.

[0192] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present disclosure. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0193] Although the embodiments of the present disclosure have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present disclosure. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present disclosure.

Claims

1. A speech signal processing method, characterized in that: The method comprises: Acquiring a speech observation signal collected by a speech collection device; performing pre-separation processing on the speech observation signal to obtain a first pre-separation signal and a second pre-separation signal, wherein a first distance between a sound source of the first pre-separation signal and the speech collection device is different from a second distance between a sound source of the second pre-separation signal and the speech collection device; and Performing blind source separation on the speech observation signal according to the first pre-separation signal to obtain a first source speech signal of the sound source of the first pre-separation signal; Blind source separation is performed on the speech observation signal according to the second pre-separation signal to obtain a second source speech signal of the sound source of the second pre-separation signal.

2. The method according to claim 1, characterized in that The speech collection device comprises a plurality of speech collection units, and the speech observation signal comprises at least: a first speech observation signal collected by a first speech collection unit, the first speech collection unit belonging to the plurality of speech collection units; The pre-separation processing of the speech observation signal to obtain a first pre-separation signal and a second pre-separation signal includes: The first speech observation signal is pre-separated to obtain a first pre-separated signal and a second pre-separated signal.

3. The method according to claim 2, characterized in that The method further comprises: A voice collection unit is randomly selected from the multiple voice collection units as the first voice collection unit.

4. The method according to claim 2, characterized in that The pre-separation processing of the first speech observation signal to obtain a first pre-separation signal and a second pre-separation signal includes: Inputting the first speech observation signal into a pre-separation model to obtain the first pre-separation signal and the second pre-separation signal output by the pre-separation model; In which, the pre-separation model is obtained by deep learning training using a training set, the training set comes from multiple speech acquisition units, the training set includes multiple samples, one speech acquisition unit corresponds to at least one sample, and each sample includes: a sample observation signal collected by the speech acquisition unit, and a first sample speech signal and a second sample speech signal corresponding to the sample observation signal, the third distance between the sound source of the first sample speech signal and the speech acquisition unit and the fourth distance between the sound source of the second sample speech signal and the speech acquisition unit are different.

5. The method according to claim 1, characterized in that The step of performing blind source separation on the speech observation signal according to the first pre-separation signal to obtain a first source speech signal of a sound source of the first pre-separation signal includes: Determining a probability density function variance term of a sound source corresponding to the speech observation signal; Using the first pre-separated signal as a pilot signal of a probability density function variance term of the sound source to obtain a probability density function variance term of the sound source into which the pilot signal is introduced; Performing blind source separation on the speech observation signal according to the first separation matrix to obtain an initial separation signal frequency vector; Determining a first separation signal frequency vector according to the initial separation signal frequency vector, the probability density function variance term of the sound source introduced into the pilot signal, and the first separation matrix; and The first source speech signal is determined according to the first separated signal frequency vector.

6. The method according to claim 5, characterized in that The determining of the first separation signal frequency vector according to the initial separation signal frequency vector, the probability density function variance term of the sound source introduced into the pilot signal, and the first separation matrix comprises: If the initial separation signal frequency vector meets the preset condition, using the initial separation signal frequency vector as the first separation signal frequency vector; If the initial separation signal frequency vector does not meet the preset condition, updating the reference item according to the first separation matrix, and obtaining the updated reference item, wherein the reference item includes: a probability density function variance item of the sound source of the pilot signal, and the first separation matrix is ​​related to the initial reference item; A second separation matrix is ​​determined according to the updated reference item, and blind source separation is performed on the speech observation signal according to the second separation matrix until the separation signal frequency vector obtained by blind source separation meets the preset condition, and the obtained separation signal frequency vector is used as the first separation signal frequency vector.

7. The method according to claim 1, characterized in that The step of performing blind source separation on the speech observation signal according to the second pre-separation signal to obtain a second source speech signal of the sound source of the second pre-separation signal comprises: Determining a probability density function variance term of a sound source corresponding to the speech observation signal; Using the second pre-separated signal as a pilot signal of a probability density function variance term of the sound source to obtain a probability density function variance term of the sound source into which the pilot signal is introduced; Performing blind source separation on the speech observation signal according to the first separation matrix to obtain an initial separation signal frequency vector; Determining a second separation signal frequency vector according to the initial separation signal frequency vector, the probability density function variance term of the sound source introduced into the pilot signal, and the first separation matrix; and The second source speech signal is determined according to the second separated signal frequency vector.

8. The method according to claim 7, characterized in that The determining of the second separation signal frequency vector according to the initial separation signal frequency vector, the probability density function variance term of the sound source introduced into the pilot signal, and the first separation matrix comprises: If the initial separation signal frequency vector meets the preset condition, using the initial separation signal frequency vector as the second separation signal frequency vector; If the initial separation signal frequency vector does not meet the preset condition, updating the reference item according to the first separation matrix, and obtaining the updated reference item, wherein the reference item includes: a probability density function variance item of the sound source of the pilot signal, and the first separation matrix is ​​related to the initial reference item; A second separation matrix is ​​determined according to the updated reference item, and blind source separation is performed on the speech observation signal according to the second separation matrix until the separation signal frequency vector obtained by blind source separation meets the preset condition, and the obtained separation signal frequency vector is used as the second separation signal frequency vector.

9. A speech signal processing device, characterized in that: The device comprises: An acquisition module, used to acquire a speech observation signal collected by a speech acquisition device; a first separation module, configured to perform pre-separation processing on the speech observation signal to obtain a first pre-separation signal and a second pre-separation signal, wherein a first distance between a sound source of the first pre-separation signal and the speech collection device is different from a second distance between a sound source of the second pre-separation signal and the speech collection device; and The second separation module is used to perform blind source separation on the speech observation signal according to the first pre-separation signal to obtain a first source speech signal of the sound source of the first pre-separation signal, and to perform blind source separation on the speech observation signal according to the second pre-separation signal to obtain a second source speech signal of the sound source of the second pre-separation signal.

10. An electronic device, characterized in that: include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 8.

11. An earphone, characterized in that: include: processor; a memory for storing processor-executable instructions; Wherein, the processor is configured to: Implement the steps of the method according to any one of claims 1 to 8.

12. A hearing aid, characterized in that: include: processor; a memory for storing processor-executable instructions; Wherein, the processor is configured to: Implement the steps of the method according to any one of claims 1 to 8.

13. A car, characterized in that: include: processor; a memory for storing processor-executable instructions; Wherein, the processor is configured to: Implement the steps of the method according to any one of claims 1 to 8.

14. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: in, The computer instructions are used to cause the computer to execute the method according to any one of claims 1 to 8.

15. A computer program product, characterized in that The invention comprises a computer program which, when executed by a processor, implements the steps of the method according to any one of claims 1 to 8.

Images