Post-processing method and device for improving voice listening feeling, equipment and medium

Abstract

The invention discloses a post-processing method and device for improving voice listening feeling, equipment and a medium. The degree of contrast stretching can be limited through the pre-configured maximum contrast stretching value, and signal distortion caused by excessive stretching is prevented. When the target contrast stretching parameter of any frequency is determined, the target contrast stretching parameter is determined by dividing the preset contrast stretching preset parameter corresponding to the frequency by the preset contrast stretching maximum value, so that the energy of the human ear non-sensitive frequency can be effectively reduced, and the energy of the human ear sensitive frequency can be kept. And subsequently, stretching processing is performed on the amplitude spectrum of the current voice frame by applying the target contrast stretching parameters corresponding to the frequencies so as to obtain a stretched amplitude spectrum, and the stretched amplitude spectrum maintains the energy of human ear sensitive frequencies and reduces the energy of non-sensitive frequencies on the basis of maintaining the original voice signal characteristics, so that the accuracy of the speech signal quality is improved. And the problem of numerical value overflow caused by excessive stretching of the amplitude of the human ear sensitive frequency is avoided.

Description

Technical Field

[0001] The present application relates to the field of digital signal processing and deep learning technology, and in particular to a post-processing method, device, equipment and medium for improving speech listening experience. Background Art

[0002] In the field of speech signal processing, post-processing is crucial to improving speech quality and intelligibility. Among them, the contrast stretching method based on the auditory characteristics of the human ear is an effective speech enhancement technology. This method deeply simulates the complex perception mechanism of the human ear to sound signals. It stretches the amplitude spectrum of the speech signal nonlinearly, especially stretching the frequency components that are more sensitive to human ear perception, while keeping the frequency components that are not sensitive to perception relatively small, thereby effectively enhancing the key features in the speech and making the processed speech clearer and more natural.

[0003] However, there is a significant problem with existing contrast stretching methods, that is, the values ​​of the stretched amplitude spectrum are unbounded and usually too large. As a result, during the processing, the values ​​of the stretched amplitude spectrum can easily exceed the maximum value of the PCM (pulse code modulation) encoding, thus causing overflow risks. Numerical overflow not only causes distortion of the voice signal, but may also damage the stability of the entire communication system, especially in real-time voice communication, real-time speech recognition and other application scenarios that require fast response and high stability. This problem is particularly prominent.

[0004] Therefore, how to maintain the contrast stretching method to effectively enhance the speech signal while quickly and effectively avoiding the numerical overflow problem has become a key technical problem that needs to be urgently solved in the current field of speech signal processing. Summary of the invention

[0005] The present application provides a post-processing method, device, equipment and medium for improving the listening experience of speech, which is used to solve the problem that the existing contrast stretching method based on human hearing cannot maintain effective enhancement of speech signals while avoiding numerical overflow.

[0006] In a first aspect, the present application provides a post-processing method for improving speech listening experience, the method comprising:

[0007] For any collected speech frame, obtaining a phase spectrum and an amplitude spectrum of the speech frame;

[0008] Obtaining target contrast stretching parameters corresponding to each frequency in the speech frame, so as to reduce the energy of frequencies insensitive to human ears through the target contrast stretching parameters corresponding to each frequency, and maintain the energy of frequencies sensitive to human ears; wherein the target contrast stretching parameter of any frequency is determined by dividing the contrast stretching preset parameter corresponding to the frequency by a pre-configured contrast stretching maximum value; the contrast stretching preset parameters corresponding to different frequencies are positively correlated with the auditory sensitivity of the human ears corresponding to the different frequencies, and the contrast stretching preset parameters corresponding to the different frequencies are all greater than 1;

[0009] By applying each of the target contrast stretching parameters, the amplitude spectrum is stretched to obtain a stretched amplitude spectrum;

[0010] synthesizing the phase spectrum and the stretched amplitude spectrum to obtain a stretched frequency spectrum;

[0011] Based on the stretched frequency spectrum, a processed speech signal is obtained.

[0012] In a second aspect, the present application further provides a post-processing device for improving speech listening experience, the device comprising:

[0013] An acquisition unit, used for acquiring a phase spectrum and an amplitude spectrum of any collected speech frame;

[0014] A parameter determination unit, used for obtaining target contrast stretching parameters corresponding to each frequency in the speech frame, so as to reduce the energy of frequencies insensitive to human ears through the target contrast stretching parameters corresponding to each frequency, and maintain the energy of frequencies sensitive to human ears; wherein the target contrast stretching parameter of any frequency is determined by dividing the contrast stretching preset parameter corresponding to the frequency by a pre-configured contrast stretching maximum value; the contrast stretching preset parameters corresponding to different frequencies are positively correlated with the auditory sensitivity of the human ears corresponding to the different frequencies, and the contrast stretching preset parameters corresponding to the different frequencies are all greater than 1;

[0015] an amplitude stretching unit, configured to stretch the amplitude spectrum by applying each of the target contrast stretching parameters to obtain a stretched amplitude spectrum;

[0016] A synthesis unit, used for synthesizing the phase spectrum and the stretched amplitude spectrum to obtain a stretched frequency spectrum;

[0017] The output unit is used to obtain a processed speech signal based on the stretched spectrum.

[0018] In a third aspect, the present application provides a computer device, the computer device comprising a processor, the processor being used to implement the steps of a post-processing method for improving speech listening experience as described above when executing a computer program stored in a memory.

[0019] In a fourth aspect, the present application provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of a post-processing method for improving speech listening experience as described above.

[0020] The beneficial effects of this application are as follows:

[0021] 1. Determine the maximum value of contrast stretching from the pre-configured contrast stretching preset parameters corresponding to different frequencies. The maximum value of contrast stretching can be used to limit the degree of contrast stretching to prevent signal distortion caused by excessive stretching.

[0022] 2. When determining the target contrast stretching parameter of any frequency, the preset contrast stretching parameter corresponding to the frequency is divided by the predetermined contrast stretching maximum value, which can effectively reduce the energy of the frequencies that are insensitive to the human ear and maintain the energy of the frequencies that are sensitive to the human ear. Subsequently, the target contrast stretching parameters corresponding to each frequency are applied to stretch the amplitude spectrum of the current speech frame to obtain a stretched amplitude spectrum. The stretched amplitude spectrum maintains the energy of the frequencies that are sensitive to the human ear while maintaining the characteristics of the original speech signal, while reducing the energy of the insensitive frequencies, thereby avoiding excessive stretching of the amplitude of the frequencies that are sensitive to the human ear and causing numerical overflow. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0024] Figure 1 A schematic diagram of a post-processing method for improving speech listening experience provided in an embodiment of the present application;

[0025] Figure 2 A flowchart of a specific post-processing method for improving speech listening experience provided in an embodiment of the present application;

[0026] Figure 3 A schematic diagram of the structure of a post-processing device for improving speech listening experience provided in an embodiment of the present application;

[0027] Figure 4It is a structural schematic diagram of a computer device provided in an optional embodiment of the present application. DETAILED DESCRIPTION

[0028] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0029] In order to maintain the contrast stretching method to effectively enhance the speech signal while quickly and effectively avoiding the numerical overflow problem, the present application provides a post-processing method, device, equipment and medium for improving the listening experience of speech.

[0030] Embodiment 1:

[0031] This application provides a post-processing method for improving the listening experience of speech. Figure 1 A schematic diagram of a post-processing method for improving speech listening experience provided in an embodiment of the present application, the process includes:

[0032] S101: For any collected speech frame, obtain the phase spectrum and amplitude spectrum of the speech frame.

[0033] The post-processing method for improving the listening experience of speech provided in the present application is applied to a computer device, which may be an intelligent terminal, such as a computer, a mobile phone, etc., or a server, such as a business server, an application server, etc.

[0034] In the present application, a computer device can collect continuous voice signals in a working environment in real time, or can obtain continuous voice signals collected or stored in real time by other devices communicating with it. Then, the voice frame contained in the continuous voice signal is obtained. Among them, the voice frame is the basic unit in voice signal processing, and usually contains a voice signal within a period of time. Exemplarily, a preset framing algorithm, such as framing and windowing processing, can be used to segment a single voice frame from the continuous voice signal. Specifically, the window length is the duration of each voice frame (or "frame length" (frame_length)), and the frame shift (frame_shift) is the step length. According to the time sequence of the voice frames, the window is moved on the continuous voice signal to obtain a voice frame sequence of the continuous voice signal. Among them, the voice frame sequence includes multiple voice frames arranged in time order. Among them, the frame shift refers to the time difference between the starting positions of two adjacent voice frames.

[0035] For any speech frame obtained, the speech frame is processed to obtain the amplitude spectrum and phase spectrum of the speech frame. Among them, the phase spectrum describes the phase information of each frequency component in the speech frame, and the amplitude spectrum describes the amplitude (or energy) information of each frequency component in the speech frame. Exemplarily, the speech frame can be first converted from the time domain to the frequency domain to obtain the spectrum of the speech frame. For example, the speech frame is converted from the time domain to the frequency domain using Fourier transform (such as short-time Fourier transform STFT) to obtain its complex spectrum. Then the spectrum is further decomposed to obtain the phase spectrum and amplitude spectrum of the speech frame. For example, the norm of the spectrum is determined as the amplitude spectrum, and the phase angle of the spectrum is determined as the phase spectrum.

[0036] For example, the phase spectrum and amplitude spectrum of a speech frame can be determined by the following formula:

[0037] A(t,f)=||X(t,f)||1

[0038] θ(t,f)=angle(X(t,f))

[0039] Where A(t,f) represents the original amplitude value of the t-th speech frame at frequency f, ||||1 represents the norm, X(t,f) represents the spectrum of the t-th speech frame at frequency f, θ(t,f) represents the phase angle of the t-th speech frame at frequency f, and angle(·) is the phase operator.

[0040] S102: Obtain target contrast stretching parameters corresponding to each frequency in the speech frame, so as to reduce the energy of frequencies insensitive to the human ear and maintain the energy of frequencies sensitive to the human ear through the target contrast stretching parameters corresponding to each frequency; wherein the target contrast stretching parameter of any frequency is determined by dividing the contrast stretching preset parameter corresponding to the frequency by a pre-configured contrast stretching maximum value; the contrast stretching preset parameters corresponding to different frequencies are positively correlated with the auditory sensitivity of the human ear corresponding to the different frequencies, and the contrast stretching preset parameters corresponding to different frequencies are all greater than 1.

[0041] In the present application, contrast stretch preset parameters corresponding to different frequencies are pre-configured. Among them, the contrast stretch preset parameters corresponding to different frequencies are intended to adjust the contrast of the voice signal at different frequencies, and the contrast here refers to the difference in strength or dynamic range of the voice signal at different frequencies. When setting the contrast stretch preset parameters corresponding to different frequencies, the perception characteristics of the human ear to sounds of different frequencies, that is, the auditory sensitivity of the human ear, can be fully considered. The auditory sensitivity of the human ear can be determined based on the auditory parameters of the human ear (such as equal loudness curves, masking effects, etc.). Specifically, because the human ear is more sensitive to mid-frequency sounds and relatively less sensitive to low-frequency and high-frequency sounds, a larger contrast stretch preset parameter can be set for the mid-frequency range to better highlight the voice features within these frequency ranges; and for the low-frequency and high-frequency ranges, a relatively small contrast stretch preset parameter is set. In other words, the setting of the contrast stretch preset parameter is positively correlated with the auditory sensitivity of the human ear.

[0042] It should be noted that the pre-configured contrast stretching preset parameters corresponding to different frequencies are all greater than 1 to ensure that the speech signal is enhanced during the stretching process.

[0043] For example, based on a 16kHz sampling rate, the preset parameters for contrast stretching of a 512-point Fourier transform (expressed as PCS) are:

[0044] PCS[0:3]=1

[0045] PCS[3:6]=1.070175439

[0046] PCS[6:9]=1.182456140

[0047] PCS[9:12]=1.287719298

[0048] PCS[12:138]=1.4

[0049] PCS[138:166]=1.322807018

[0050] PCS[166:200]=1.238596491

[0051] PCS[200:241]=1.161403509

[0052] PCS[241:256]=1.077192982

[0053] At the same time, since the pre-configured contrast stretching preset parameters corresponding to different frequencies are all greater than 1, the purpose is to increase the energy of the sensitive area of ​​the human ear, but directly based on the contrast stretching preset parameters, after stretching the amplitude spectrum, the stretched amplitude spectrum value will be unbounded and it is easy to exceed the maximum value of the PCM (pulse code modulation) encoding, thereby causing overflow risk. Therefore, in this application, a maximum value (recorded as the contrast stretching maximum value) can be determined from the pre-configured contrast stretching preset parameters corresponding to different frequencies to limit the degree of contrast stretching and prevent signal distortion caused by excessive stretching.

[0054] After the amplitude spectrum of the speech frame is obtained based on the above embodiment, the target contrast stretching parameters corresponding to each frequency in the speech frame can be obtained, so as to reduce the energy of the frequency insensitive to the human ear through the target contrast stretching parameters corresponding to each frequency, and maintain the energy of the frequency sensitive to the human ear. When determining the target contrast stretching parameter of any frequency, the preset contrast stretching preset parameter corresponding to the frequency can be divided by the predetermined contrast stretching maximum value.

[0055] For example, the target contrast stretch parameter for any frequency can be determined by the following formula:

[0056] PCS fix (f) = PCS(f) / max(PCS)

[0057] Among them, PCS fix (f) represents the target contrast stretching parameter corresponding to frequency f, PCS(f) represents the contrast stretching preset parameter corresponding to frequency f, and max(PCS) represents the maximum contrast stretching value.

[0058] It should be noted that there are two ways to determine the target contrast stretching parameters:

[0059] 1. Predetermination: In this way, before processing the speech frame, the target contrast stretching parameters corresponding to different frequencies can be pre-calculated based on the contrast stretching maximum value and the contrast stretching preset parameters corresponding to different frequencies, and stored for subsequent use.

[0060] 2. Real-time determination: Another way is to calculate the target contrast stretching parameters corresponding to each frequency in the current speech frame in real time when processing each speech frame based on the contrast stretching preset parameters corresponding to each frequency in the current speech frame and the pre-configured contrast stretching maximum value.

[0061] When a certain speech frame is processed, according to its frequency components, the contrast stretch preset parameters corresponding to each frequency in the speech frame are searched from the pre-configured contrast stretch preset parameters corresponding to different frequencies.

[0062] S103: Performing stretching processing on the amplitude spectrum by applying each of the target contrast stretching parameters to obtain a stretched amplitude spectrum.

[0063] After obtaining the target contrast stretching parameters corresponding to each frequency in the speech frame based on the above embodiment, the amplitude spectrum of the speech frame can be stretched by the target contrast stretching parameters corresponding to each frequency component. Exemplarily, it can be achieved by multiplying each frequency component of the amplitude spectrum by the corresponding target contrast stretching parameter.

[0064] In a possible implementation, for each frequency in the speech frame, the amplitude corresponding to the frequency can be obtained in the amplitude spectrum of the speech frame. The sum of the amplitude and 1 is determined to avoid the situation where 0 or a negative number appears when taking the logarithm. The natural logarithm of this sum is then calculated to convert the amplitude value to a range that is more suitable for stretching processing. The obtained natural logarithm value is multiplied by the target contrast stretching parameter corresponding to the frequency. The obtained product is added to 1 to maintain the non-zero nature of the amplitude. The natural exponent of the sum of the product and 1 is then calculated to obtain a natural exponent value. The natural exponent value is determined as the amplitude value after stretching.

[0065] For example, by applying each of the target contrast stretching parameters, the amplitude spectrum is stretched to obtain a stretched amplitude spectrum, which is expressed by the following formula:

[0066]

[0067] Among them, A P (t,f) represents the amplitude value of the t-th speech frame after stretching at frequency f. PCS fix (f) represents the target contrast stretching parameter corresponding to frequency f, A(t,f) represents the original amplitude value of the t-th speech frame at frequency f, and e (·) is the natural exponential operator, Ln (·) is the natural logarithm operator.

[0068] By repeating the above steps, a stretched amplitude value can be calculated for each frequency in the speech frame. By combining these values, a stretched amplitude spectrum is obtained. The stretched amplitude spectrum maintains the energy of the frequencies sensitive to the human ear while reducing the energy of the insensitive frequencies on the basis of maintaining the original speech signal characteristics, thus avoiding the problem of numerical overflow caused by excessive stretching of the amplitudes of the frequencies sensitive to the human ear.

[0069] S104: synthesizing the phase spectrum and the stretched amplitude spectrum to obtain a stretched frequency spectrum.

[0070] After the stretched amplitude spectrum is obtained based on the above embodiment, the stretched amplitude spectrum can be synthesized with the original phase spectrum to obtain the stretched frequency spectrum.

[0071] Exemplarily, for each frequency in the speech frame, the stretched amplitude corresponding to the frequency is used as the modulus of the complex number, and the phase corresponding to the frequency in the original phase spectrum is used as the argument of the complex number. Then, the conversion formula between polar coordinates and rectangular coordinates (i.e., Euler's formula) is used to convert this modulus and argument into a complex number. The complex numbers corresponding to all frequencies are combined to obtain the stretched spectrum. This spectrum contains both the stretched amplitude information and the original phase information.

[0072] In another exemplary embodiment, for each frequency in the speech frame, the phase angle corresponding to the frequency can be obtained in the phase spectrum of the speech frame. Then the product of the phase angle and the imaginary unit is determined. The imaginary unit is equal to negative one under the square root. The natural exponent of the product is calculated to obtain a natural exponent value. Then, the natural exponent value is multiplied by the stretched amplitude value corresponding to the frequency, and the obtained product is determined as the stretched spectrum.

[0073] For example,

[0074] X p (t,f)=A P (t,f)·e jθ(t,f)

[0075] Among them, X p (t,f) represents the stretched spectrum of the tth speech frame at frequency f, A P (t,f) represents the amplitude value of the t-th speech frame after stretching at frequency f, θ(t,f) represents the phase angle of the t-th speech frame at frequency f, j represents the imaginary unit, e (·) is the natural exponential operator.

[0076] S105: Obtaining a processed speech signal based on the stretched frequency spectrum.

[0077] Finally, after obtaining the stretched spectrum based on the above embodiment, the stretched spectrum can be converted from the frequency domain back to the time domain using an inverse Fourier transform (such as an inverse short-time Fourier transform (ISTFT)) to obtain a processed speech signal. The processed speech signal has higher clarity and naturalness while maintaining the original speech content, especially within the frequency range sensitive to the human ear.

[0078] In a possible implementation, the spectrum after contrast stretching has been adjusted in the frequency domain. In order to reflect these adjustments to the time domain signal, the stretched spectrum can be transformed using an inverse Fourier transform (IFT), thereby converting the frequency domain signal back to the time domain, thereby obtaining a processed speech frame time domain signal.

[0079] Since speech signals are usually continuous and long, they are usually divided into multiple shorter speech frames for processing. However, this segmentation may cause discontinuity between frames, thereby affecting the quality of the processed speech. In order to solve this problem, when segmenting the speech signal, adjacent frames can be overlapped. The length of the overlap usually depends on the window function used (such as Hanning window, Hamming window, etc.) and the specific signal processing task. After the processed time domain signals corresponding to each speech frame are obtained, the processing results of adjacent frames can be added according to a certain weight in the overlapping area, for example, the value of the window function is used as the weight to smooth the transition and reduce the discontinuity between adjacent frames. The processed speech frames (including the overlapping parts) are spliced ​​together to form a complete processed speech signal. Through the overlap-add method, the discontinuity between frames can be effectively reduced and the quality of the processed speech can be improved.

[0080] The beneficial effects of this application are as follows:

[0081] 1. Determine the maximum value of contrast stretching from the pre-configured contrast stretching preset parameters corresponding to different frequencies. The maximum value of contrast stretching can be used to limit the degree of contrast stretching to prevent signal distortion caused by excessive stretching.

[0082] 2. When determining the target contrast stretching parameter of any frequency, the preset contrast stretching parameter corresponding to the frequency is divided by the predetermined contrast stretching maximum value, which can effectively reduce the energy of the frequencies that are insensitive to the human ear and maintain the energy of the frequencies that are sensitive to the human ear. Subsequently, the target contrast stretching parameters corresponding to each frequency are applied to stretch the amplitude spectrum of the current speech frame to obtain a stretched amplitude spectrum. The stretched amplitude spectrum maintains the energy of the frequencies that are sensitive to the human ear while maintaining the characteristics of the original speech signal, while reducing the energy of the insensitive frequencies, thereby avoiding excessive stretching of the amplitude of the frequencies that are sensitive to the human ear and causing numerical overflow.

[0083] Embodiment 2:

[0084] The following is a description of a post-processing method for improving speech listening provided by the present application through a specific embodiment. Figure 2A schematic flow chart of a specific post-processing method for improving speech listening experience provided in an embodiment of the present application, the process includes:

[0085] S201: Acquire a continuous speech signal.

[0086] S202: performing frame division and windowing processing on the continuous speech signal to obtain speech frames in the continuous speech signal.

[0087] For each speech frame in the continuous speech signal, the following S203 to S208 are performed:

[0088] S203: Perform fast Fourier transform on the speech frame to obtain a frequency spectrum of the speech frame.

[0089] S204: Determine the amplitude spectrum and phase spectrum of the speech frame based on the frequency spectrum.

[0090] S205: Obtain target contrast stretching parameters corresponding to each frequency in the speech frame.

[0091] The target contrast stretch parameter of any frequency is determined by dividing the preset contrast stretch parameter corresponding to the frequency by the pre-configured maximum contrast stretch value. The preset contrast stretch parameters corresponding to different frequencies are positively correlated with the auditory sensitivity of the human ear corresponding to different frequencies, and the preset contrast stretch parameters corresponding to different frequencies are all greater than 1.

[0092] S206: Perform stretching processing on the amplitude spectrum by applying each target contrast stretching parameter to obtain a stretched amplitude spectrum.

[0093] In one possible implementation, for each frequency in the speech frame, the amplitude corresponding to the frequency is obtained from the amplitude spectrum; the natural logarithm of the sum of the amplitude and 1 is determined to obtain a natural logarithm value; the product of the target contrast stretching parameter and the natural logarithm value is determined; and based on the natural exponent of the sum of the product and 1, the stretched amplitude value is determined.

[0094] S207: Synthesize the phase spectrum and the stretched amplitude spectrum to obtain a stretched frequency spectrum.

[0095] In one possible implementation, for each frequency in the speech frame, the phase angle corresponding to the frequency is obtained from the phase spectrum; the natural exponent of the product of the phase angle and the imaginary unit is determined to obtain a natural exponent value; wherein the imaginary unit is equal to negative one under the square root; and the stretched frequency spectrum is determined based on the product of the natural exponent value and the stretched amplitude value corresponding to the frequency.

[0096] S208: Perform inverse Fourier transform on the stretched spectrum to obtain a processed time domain signal of the speech frame.

[0097] S209: Processing the processed time domain signals corresponding to the respective speech frames by overlap-add method to obtain processed speech signals.

[0098] Embodiment 3:

[0099] Based on the same inventive concept, the present application also provides a post-processing device for improving the sense of hearing of speech. Figure 3 A schematic diagram of the structure of a post-processing device for improving speech listening experience provided in an embodiment of the present application, the device comprising:

[0100] An acquisition unit 31 is used to acquire a phase spectrum and an amplitude spectrum of any collected speech frame;

[0101] The parameter determination unit 32 is used to obtain the target contrast stretching parameters corresponding to each frequency in the speech frame, so as to reduce the energy of the frequency insensitive to the human ear through the target contrast stretching parameters corresponding to each frequency, and maintain the energy of the frequency sensitive to the human ear; wherein the target contrast stretching parameter of any frequency is determined by dividing the contrast stretching preset parameter corresponding to the frequency by the pre-configured contrast stretching maximum value; the contrast stretching preset parameters corresponding to different frequencies are positively correlated with the auditory sensitivity of the human ear corresponding to the different frequencies, and the contrast stretching preset parameters corresponding to the different frequencies are all greater than 1;

[0102] an amplitude stretching unit 33, configured to perform stretching processing on the amplitude spectrum by applying each of the target contrast stretching parameters to obtain a stretched amplitude spectrum;

[0103] A synthesis unit 34, used for synthesizing the phase spectrum and the stretched amplitude spectrum to obtain a stretched frequency spectrum;

[0104] The output unit 35 is used to obtain a processed speech signal based on the stretched spectrum.

[0105] In some possible implementations, the amplitude stretching unit 33 is specifically used to:

[0106] For each frequency, obtain the amplitude corresponding to the frequency from the amplitude spectrum; determine the natural logarithm of the sum of the amplitude and 1 to obtain a natural logarithm value; determine the product of the target contrast stretching parameter and the natural logarithm value; and determine the amplitude value after stretching based on the natural exponent of the sum of the product and 1.

[0107] In some possible implementations, the synthesis unit 34 is specifically used to:

[0108] For each frequency, obtain the phase angle corresponding to the frequency from the phase spectrum; determine the natural exponent of the product of the phase angle and an imaginary unit to obtain a natural exponent value; wherein the imaginary unit is equal to negative one under the square root; and determine the stretched spectrum based on the product of the natural exponent value and the stretched amplitude value corresponding to the frequency.

[0109] In some possible implementations, the acquisition unit 31 is specifically configured to:

[0110] Acquire the frequency spectrum of the speech frame;

[0111] The norm of the frequency spectrum is determined as the magnitude spectrum, and the phase angle of the frequency spectrum is determined as the phase spectrum.

[0112] In some possible implementations, the acquisition unit 31 is specifically configured to:

[0113] The collected continuous speech signal is framed and windowed to obtain speech frames in the continuous speech signal.

[0114] In some possible implementations, the output unit 35 is specifically used to:

[0115] Performing an inverse Fourier transform on the stretched spectrum to obtain a processed time domain signal of the speech frame;

[0116] The processed time domain signals corresponding to the speech frames are processed by overlap-addition method to obtain processed speech signals.

[0117] In this embodiment, a post-processing device for improving the listening experience of speech is presented in the form of a functional module, where the module refers to an application specific integrated circuit (ASIC), a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0118] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0119] Embodiment 4:

[0120] See also Figure 4 , Figure 4 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present application, such as Figure 4As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 4 A processor 10 is taken as an example.

[0121] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0122] The memory 20 stores instructions executable by at least one processor 10, so that at least one processor 10 executes the method shown in the above embodiment.

[0123] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created by the use of a computer device based on the presentation of a small program landing page, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0124] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0125] The computer device also includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30 and the output device 40 may be connected via a bus or other means. Figure 4 The example of connecting through bus is taken in the following.

[0126] The input device 30 can receive input digital or character information, and generate key signal input related to the user settings and function control of the computer device, such as a touch screen, a keypad, a mouse, a track pad, a touch pad, an indicator bar, one or more mouse buttons, a trackball, a joystick, etc. The output device 40 may include a display device, an auxiliary lighting device (e.g., an LED) and a tactile feedback device (e.g., a vibration motor), etc. The above-mentioned display device includes but is not limited to a liquid crystal display, a light emitting diode, a display and a plasma display. In some optional embodiments, the display device can be a touch screen.

[0127] Embodiment 5:

[0128] On the basis of the above embodiments, an embodiment of the present application further provides a computer-readable storage medium, in which a computer program executable by a processor is stored. When the program runs on the processor, the processor implements the following steps when executing:

[0129] For any collected speech frame, obtaining a phase spectrum and an amplitude spectrum of the speech frame;

[0130] Obtaining target contrast stretching parameters corresponding to each frequency in the speech frame, so as to reduce the energy of frequencies insensitive to human ears through the target contrast stretching parameters corresponding to each frequency, and maintain the energy of frequencies sensitive to human ears; wherein the target contrast stretching parameter of any frequency is determined by dividing the contrast stretching preset parameter corresponding to the frequency by a pre-configured contrast stretching maximum value; the contrast stretching preset parameters corresponding to different frequencies are positively correlated with the auditory sensitivity of the human ears corresponding to the different frequencies, and the contrast stretching preset parameters corresponding to the different frequencies are all greater than 1;

[0131] By applying each of the target contrast stretching parameters, the amplitude spectrum is stretched to obtain a stretched amplitude spectrum;

[0132] synthesizing the phase spectrum and the stretched amplitude spectrum to obtain a stretched frequency spectrum;

[0133] Based on the stretched frequency spectrum, a processed speech signal is obtained.

[0134] Since the principle of solving the problem by the above-mentioned computer-readable storage medium is similar to a post-processing method for improving speech listening experience, the implementation of the above-mentioned computer-readable storage medium can refer to the embodiment of the method, and the repeated parts will not be repeated.

[0135] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A post-processing method for improving speech listening quality, characterized in that: The method comprises: For any collected speech frame, obtaining a phase spectrum and an amplitude spectrum of the speech frame; Obtaining target contrast stretching parameters corresponding to each frequency in the speech frame, so as to reduce the energy of frequencies insensitive to human ears through the target contrast stretching parameters corresponding to each frequency, and maintain the energy of frequencies sensitive to human ears; wherein the target contrast stretching parameter of any frequency is determined by dividing the contrast stretching preset parameter corresponding to the frequency by a pre-configured contrast stretching maximum value; the contrast stretching preset parameters corresponding to different frequencies are positively correlated with the auditory sensitivity of the human ears corresponding to the different frequencies, and the contrast stretching preset parameters corresponding to the different frequencies are all greater than 1; By applying each of the target contrast stretching parameters, the amplitude spectrum is stretched to obtain a stretched amplitude spectrum; synthesizing the phase spectrum and the stretched amplitude spectrum to obtain a stretched frequency spectrum; Based on the stretched frequency spectrum, a processed speech signal is obtained.

2. The method according to claim 1, characterized in that The stretching process is performed on the amplitude spectrum by applying each of the target contrast stretching parameters to obtain a stretched amplitude spectrum, including: For each frequency, obtain the amplitude corresponding to the frequency from the amplitude spectrum; determine the natural logarithm of the sum of the amplitude and 1 to obtain a natural logarithm value; determine the product of the target contrast stretching parameter and the natural logarithm value; and determine the amplitude value after stretching based on the natural exponent of the sum of the product and 1.

3. The method according to claim 1, characterized in that The step of synthesizing the phase spectrum with the stretched amplitude spectrum to obtain a stretched frequency spectrum includes: For each frequency, obtain the phase angle corresponding to the frequency from the phase spectrum; determine the natural exponent of the product of the phase angle and an imaginary unit to obtain a natural exponent value; wherein the imaginary unit is equal to negative one under the square root; and determine the stretched spectrum based on the product of the natural exponent value and the stretched amplitude value corresponding to the frequency.

4. The method according to claim 1, characterized in that For any collected speech frame, obtaining the phase spectrum and amplitude spectrum of the speech frame includes: Acquire the frequency spectrum of the speech frame; The norm of the frequency spectrum is determined as the magnitude spectrum, and the phase angle of the frequency spectrum is determined as the phase spectrum.

5. The method according to claim 1, characterized in that Acquiring the voice frame includes: The collected continuous speech signal is framed and windowed to obtain speech frames in the continuous speech signal.

6. The method according to claim 5, characterized in that The step of obtaining a processed speech signal based on the stretched spectrum includes: Performing an inverse Fourier transform on the stretched spectrum to obtain a processed time domain signal of the speech frame; The processed time domain signals corresponding to the speech frames are processed by overlap-addition method to obtain processed speech signals.

7. A post-processing device for improving speech listening quality, characterized in that: The device comprises: An acquisition unit, used for acquiring a phase spectrum and an amplitude spectrum of any collected speech frame; A parameter determination unit, used for obtaining target contrast stretching parameters corresponding to each frequency in the speech frame, so as to reduce the energy of frequencies insensitive to human ears through the target contrast stretching parameters corresponding to each frequency, and maintain the energy of frequencies sensitive to human ears; wherein the target contrast stretching parameter of any frequency is determined by dividing the contrast stretching preset parameter corresponding to the frequency by a pre-configured contrast stretching maximum value; the contrast stretching preset parameters corresponding to different frequencies are positively correlated with the auditory sensitivity of the human ears corresponding to the different frequencies, and the contrast stretching preset parameters corresponding to the different frequencies are all greater than 1; an amplitude stretching unit, configured to stretch the amplitude spectrum by applying each of the target contrast stretching parameters to obtain a stretched amplitude spectrum; A synthesis unit, used for synthesizing the phase spectrum and the stretched amplitude spectrum to obtain a stretched frequency spectrum; The output unit is used to obtain a processed speech signal based on the stretched spectrum.

8. The device according to claim 7, characterized in that The amplitude stretching unit is specifically used for: For each frequency, obtaining the amplitude corresponding to the frequency from the amplitude spectrum; determining the natural logarithm of the sum of the amplitude and 1 to obtain a natural logarithm value; determining the product of the target contrast stretch parameter and the natural logarithm value; Based on the natural exponential of the sum of the product and 1, the amplitude value after stretching is determined.

9. A computer device, characterized in that: The computer device includes a processor, and the processor is used to implement the steps of a post-processing method for improving speech hearing as described in any one of claims 1 to 6 when executing a computer program stored in a memory.

10. A computer-readable storage medium, characterized in that: It stores a computer program, and when the computer program is executed by a processor, the steps of a post-processing method for improving the sense of hearing of speech as described in any one of claims 1 to 6 are implemented.

Images