Voice Pre-Compensation Processing Method and Digital Power Amplifier Device

Abstract

The present application provides a voice pre-compensation processing method and a digital power amplifier device, which relate to the technical field of sound processing. The voice pre-compensation processing module inputs the obtained short-time objective intelligibility (STOI) influence value, perceptual objective listening quality assessment (POLQA) influence value, segmented signal-to-noise ratio influence value, and volume influence value into the voice adjustment parameter determination model to determine the gain adjustment parameter of the AGC module, the gain adjustment parameters of the DRC module in different time domains, the gain adjustment parameters of the equalizer in different frequency domains, and the gain adjustment parameter of the digital power amplifier module. According to the gain adjustment parameter of the AGC module, the gain control parameter configured for the AGC module is adjusted, according to the gain adjustment parameters of the DRC module in different time domains, the gain control parameter configured for the DRC module is adjusted, and according to the gain adjustment parameters of the equalizer in different frequency domains, the gain control parameter configured for the equalizer is adjusted, so that the voice played by the speaker is no longer affected by environmental erosion and has high voice quality.

Description

Technical Field

[0001] This application relates to the technical field of sound processing, and particularly to a voice pre-compensation processing method and a digital power amplifier device. Background Art

[0002] In scenarios such as hotels, villas, restaurants, conference rooms, and living rooms where the user's voice needs to be amplified, digital power amplifier devices and speakers for playing the amplified voice can be used. However, the performance of the speaker is easily affected by the environment. In a high-humidity environment, water molecules may penetrate into the speaker, causing the diaphragm to get damp. This will affect the vibration characteristics of the diaphragm, possibly resulting in sound distortion, volume reduction, or even noise. In high or low temperature environments, the performance of the speaker will be affected, resulting in volume changes or sound distortion.

[0003] Current digital power amplifier devices can convert the received analog voice signal into a digital voice signal, then use a digital signal processor (DSP) to perform noise reduction processing and amplification on the digital voice signal, and finally convert the processed digital voice signal into an analog voice signal for output to drive the speaker to play the voice. However, when the performance of the speaker is affected by the environment, the quality of the voice played by the speaker after being processed by the digital power amplifier device is low and does not meet the user's requirements. Summary of the Invention

[0004] This application provides a voice pre-compensation processing method and a digital power amplifier device, which are used to solve the problem that in the prior art, when the performance of the speaker is affected by the environment, the quality of the voice played by the speaker after being processed by the digital power amplifier device is low and does not meet the user's requirements.

[0005] In a first aspect, this application provides a voice pre-compensation processing method, which is applied to a digital power amplifier device. The digital power amplifier device includes a housing, on the surface of which a voice input interface, a voice output interface, a temperature sensor, and a humidity sensor are arranged at intervals. Inside the housing, a voice pre-compensation processing module, an automatic gain control (AGC) module, a dynamic range control (DRC) module, an equalizer, and a digital power amplifier module are arranged. The voice input interface, the AGC module, the DRC module, the equalizer, the digital power amplifier module, and the voice output interface are electrically connected in sequence. The voice input interface, the temperature sensor, the humidity sensor, the AGC module, the DRC module, and the equalizer are respectively electrically connected to the voice pre-compensation processing module. The method includes:

[0006] The voice pre-compensation processing module receives the ambient temperature data sequence collected by the temperature sensor and the ambient humidity data sequence collected by the humidity sensor;

[0007] When the voice pre-compensation processing module monitors voice data from a peripheral microphone via the voice input interface, it intercepts a sub-sequence of ambient temperature data collected within a previously preset duration from the ambient temperature data sequence, and a sub-sequence of ambient humidity data collected within a previously preset duration from the ambient humidity data sequence;

[0008] The voice pre-compensation processing module identifies the sub-sequence of ambient temperature data and the sub-sequence of ambient humidity data to determine whether they have an impact on the voice playback of the speaker connected to the voice output interface;

[0009] If there is an impact, the voice pre-compensation processing module inputs the sub-sequence of ambient temperature data, the sub-sequence of ambient temperature data, the preset simulation structure model of the speaker, and the material type of the speaker into a pre-trained voice quality parameter determination model to obtain the short-time objective intelligibility (STOI) impact value, the perceptual objective voice quality assessment (POLQA) impact value, the segmental signal-to-noise ratio impact value, and the volume impact value of the voice played by the speaker. Among them, the voice quality parameter determination model is obtained by training multiple first training samples input into a first neural network. Each first training sample includes a historical sub-sequence of ambient temperature data, a historical sub-sequence of ambient temperature data, a historical simulation structure model of the speaker, the material type of the historical speaker, and the corresponding historical actual short-time objective intelligibility (STOI) impact value, historical actual perceptual objective voice quality assessment (POLQA) impact value, historical actual segmental signal-to-noise ratio impact value, and historical actual volume impact value;

[0010] Input the short-time objective intelligibility (STOI) impact value, the perceptual objective voice quality assessment (POLQA) impact value, the segmental signal-to-noise ratio impact value, and the volume impact value into a pre-trained voice adjustment parameter determination model to determine the gain adjustment parameter of the AGC module, the gain adjustment parameters of the DRC module in different time domains, and the gain adjustment parameters of the equalizer in different frequency domains. Among them, the voice adjustment parameter determination model is obtained by training multiple second training samples input into a second neural network. Each second training sample includes a historical short-time objective intelligibility (STOI) impact value, a historical perceptual objective voice quality assessment (POLQA) impact value, a historical segmental signal-to-noise ratio impact value, a historical volume impact value, and the corresponding historical actual gain adjustment parameter of the AGC module, the historical actual gain adjustment parameters of the DRC module in different time domains, and the historical actual gain adjustment parameters of the equalizer in different frequency domains;

[0011] The voice pre-compensation processing module adjusts the gain control parameter configured in the AGC module according to the gain adjustment parameter of the AGC module, adjusts the gain control parameter configured in the DRC module according to the gain adjustment parameters of the DRC module in different time domains, and adjusts the gain control parameter configured in the equalizer according to the gain adjustment parameters of the equalizer in different frequency domains;

[0012] The voice pre - compensation processing module controls the AGC module to process voice data based on the adjusted gain control parameter of the AGC and output it to the DRC module. The DRC module processes the voice data based on the adjusted gain control parameter of the DRC module and outputs it to the equalizer. The equalizer processes the voice data based on the adjusted gain control parameter of the equalizer and outputs it to the digital power amplifier module;

[0013] The digital power amplifier module sequentially performs analog - to - digital conversion, filtering processing, amplification processing, and digital - to - analog conversion on the received voice data, and transmits it to the speaker for playback via the voice output interface.

[0014] In some embodiments, the voice pre - compensation processing module identifies the ambient temperature data subsequence and the ambient humidity data subsequence, and determines whether there is an impact on the voice playback of the speaker connected to the voice output interface, including:

[0015] The voice pre - compensation processing module inputs the ambient temperature data subsequence and the ambient humidity data subsequence into a pre - trained voice playback impact determination model to determine whether there is an impact on the voice playback of the speaker connected to the voice output interface; wherein, the voice playback impact determination model is obtained by training multiple third training samples input into a third neural network model, and each third training sample includes a historical ambient temperature data subsequence, a historical ambient humidity data subsequence, and a corresponding historical impact result label indicating whether there is an impact on the voice playback of the speaker.

[0016] In some embodiments, the voice pre - compensation processing module identifies the ambient temperature data subsequence and the ambient humidity data subsequence, and determines whether there is an impact on the voice playback of the speaker connected to the voice output interface, including:

[0017] The voice pre - compensation processing module determines the average ambient temperature within a previous preset time period according to the ambient temperature data subsequence, and determines the average ambient humidity within a previous preset time period according to the ambient humidity data subsequence;

[0018] Determine whether the average ambient temperature is greater than the set temperature threshold, and / or whether the average ambient humidity is greater than the set humidity threshold;

[0019] If the average ambient temperature is greater than the set temperature threshold, and / or the average ambient humidity is greater than the set humidity threshold, it is determined that there is an impact on the voice playback of the speaker connected to the voice output interface.

[0020] In some embodiments, a display panel is also embedded on the surface of the housing. The display panel is electrically connected to the voice pre-compensation processing module. After obtaining the short-term objective intelligibility (STOI) influence value, the perceptual objective speech quality assessment (POLQA) influence value, the segmented signal-to-noise ratio influence value, and the volume influence value of the voice played by the speaker, the method further includes:

[0021] Determine whether the speaker has a fault according to the short-term objective intelligibility (STOI) influence value, the perceptual objective speech quality assessment (POLQA) influence value, the segmented signal-to-noise ratio influence value, and the volume influence value of the voice played by the speaker;

[0022] If there is a fault, output a prompt message for indicating that the speaker has a fault to the display panel.

[0023] In some embodiments, determining whether the speaker has a fault according to the short-term objective intelligibility (STOI) influence value, the perceptual objective speech quality assessment (POLQA) influence value, the segmented signal-to-noise ratio influence value, and the volume influence value of the voice played by the speaker includes:

[0024] According to the formula P = k 1 a + k 2 b + k 3 c + k 4 d, determine the comprehensive influence factor of the voice playback of the speaker, where P is the comprehensive influence factor of the voice playback of the speaker, a is the short-term objective intelligibility (STOI) influence value, b is the perceptual objective speech quality assessment (POLQA) influence value, c is the segmented signal-to-noise ratio influence value, d is the volume influence value, and k 1 is the preset first weighting coefficient, k 2 is the preset second weighting coefficient, k 3 is the preset third weighting coefficient, k 4 is the preset fourth weighting coefficient, and k 1 + k 2 + k 3 + k 4 = 1;

[0025] When the comprehensive influence factor of the voice playback of the speaker is greater than the set influence threshold, determine that the speaker has a fault.

[0026] In a second aspect, the present application further provides a digital power amplifier device, which includes a housing. On the surface of the housing, a voice input interface, a voice output interface, a temperature sensor, and a humidity sensor are spaced apart. Inside the housing, there are a voice pre-compensation processing module, an automatic gain control (AGC) module, a dynamic range control (DRC) module, an equalizer, and a digital power amplifier module. The voice input interface, the AGC module, the DRC module, the equalizer, the digital power amplifier module, and the voice output interface are electrically connected in sequence. The voice input interface, the temperature sensor, the humidity sensor, the AGC module, the DRC module, and the equalizer are respectively electrically connected to the voice pre-compensation processing module, and the voice pre-compensation processing module is used for the voice pre-compensation processing method provided in the first aspect of the present application.

[0027] In some embodiments, a power supply port is further provided on the surface of the housing, and the power supply port is electrically connected to the voice pre-compensation processing module.

[0028] In some embodiments, the housing is a rectangular parallelepiped box.

[0029] In a third aspect, the present application further provides a storage medium storing a computer program. When the computer program is executed by a processor, the computer is made to execute the method performed by the voice pre-compensation processing module in the first aspect of the present application.

[0030] In a fourth aspect, the present application further provides a computer program product including a computer program. When the computer program is run, the computer is made to execute the method performed by the voice pre-compensation processing module in the first aspect of the present application.

[0031] The present application provides a voice pre-compensation processing method and a digital power amplifier device. When the voice pre-compensation processing module determines that the ambient temperature data subsequence and the ambient humidity data subsequence affect the voice playback of the speaker connected to the voice output interface, it inputs the ambient temperature data subsequence, the ambient temperature data subsequence, the preset simulation structure model of the speaker, and the material type of the speaker used to manufacture the speaker into the pre-trained voice quality parameter determination model to obtain the short-term objective intelligibility (STOI) influence value, the perceptual objective voice quality assessment (POLQA) influence value, the segmented signal-to-noise ratio influence value, and the volume influence value on the voice played by the speaker. Since the voice quality parameter determination model is trained by inputting multiple first training samples into the first neural network, the accuracy of the output result is high. Among them, the short-term objective intelligibility (STOI) influence value can be understood as the difference between the STOI index of the speaker playing the voice normally without being affected and the STOI index of the speaker playing the voice when being affected; the perceptual objective voice quality assessment (POLQA) influence value can be understood as the difference between the POLQA index of the speaker playing the voice normally without being affected and the POLQA index of the speaker playing the voice when being affected, the segmented signal-to-noise ratio influence value can be understood as the difference between the segmented signal-to-noise ratio of the speaker playing the voice normally without being affected and the segmented signal-to-noise ratio of the speaker playing the voice when being affected, and the volume influence value can be understood as the difference between the volume of the speaker playing the voice normally without being affected and the volume of the STOI index of the speaker playing the voice when being affected.

[0032] Input the short-term objective intelligibility (STOI) influence value, the perceptual objective voice quality assessment (POLQA) influence value, the segmented signal-to-noise ratio influence value, and the volume influence value into the pre-trained voice adjustment parameter determination model to determine the gain adjustment parameter of the AGC module, the gain adjustment parameters of the DRC module in different time domains, the gain adjustment parameters of the equalizer in different frequency domains, and the gain adjustment parameter of the digital power amplifier module. Since the voice adjustment parameter determination model is trained by inputting multiple second training samples into the second neural network, the accuracy of the output result is high;

[0033] Furthermore, the voice pre-compensation processing module adjusts the gain control parameter configured in the AGC module according to the gain adjustment parameter of the AGC module, adjusts the gain control parameter configured in the DRC module according to the gain adjustment parameters of the DRC module in different time domains, and adjusts the gain control parameter configured in the equalizer according to the gain adjustment parameters of the equalizer in different frequency domains.

[0034] The voice pre - compensation processing module controls the AGC module to process voice data based on the adjusted gain control parameters of the AGC and output it to the DRC module. The DRC module processes the voice data based on the adjusted gain control parameters of the DRC module and outputs it to the equalizer. The equalizer processes the voice data based on the adjusted gain control parameters of the equalizer and outputs it to the digital power amplifier module. The digital power amplifier module performs analog - to - digital conversion, filtering, amplification, and digital - to - analog conversion on the received voice data, and transmits it to the speaker for playback through the voice output interface. It can be understood that when the performance of the speaker is affected by the environment, since the played voice has been pre - compensated by the gain adjustment parameters of the AGC module, the gain adjustment parameters of the DRC module in different time domains, the gain adjustment parameters of the equalizer in different frequency domains, and the gain adjustment parameters of the digital power amplifier module, the STOI index, POLQA index, segmental signal - to - noise ratio, volume, and other indexes affecting voice quality of the voice can be compensated. Furthermore, the voice played by the speaker is no longer affected by the environmental erosion, and the voice quality is high, meeting the user's requirements. Brief Description of the Drawings

[0035] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0036] Figure 1 It is a circuit connection block diagram of the digital power amplifier device provided by the embodiment of the present application;

[0037] Figure 2 It is a flowchart of the voice pre - compensation processing method provided by the embodiment of the present application. Detailed Embodiments

[0038] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some, but not all, of the embodiments of the present application. Based on the embodiments of the present application, all other embodiments made by those of ordinary skill in the art under the inspiration of this embodiment belong to the scope of protection of the present application.

[0039] In the description, claims, and above-mentioned drawings of this application, terms such as "first", "second", "third", "fourth", etc. (if any) are used to distinguish similar objects and do not necessarily describe a specific order or sequence. It should be understood that such data used can be interchanged under appropriate circumstances so that the embodiments of this application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0040] An embodiment of this application provides a voice pre-compensation processing method, which is applied to a digital power amplifier device. The digital power amplifier device includes a housing, on the surface of which a voice input interface, a voice output interface, a temperature sensor, and a humidity sensor are spaced apart. Inside the housing, a voice pre-compensation processing module, an automatic gain control (AGC) module, a dynamic range control (DRC) module, an equalizer, and a digital power amplifier module are provided. As Figure 1 shown, the voice input interface, the AGC module, the DRC module, the equalizer, the digital power amplifier module, and the voice output interface are electrically connected in sequence, and the voice input interface, the temperature sensor, the humidity sensor, the AGC module, the DRC module, and the equalizer are respectively electrically connected to the voice pre-compensation processing module. As Figure 2 shown, the method provided by the embodiment of this application includes:

[0041] S201: The voice pre-compensation processing module receives the ambient temperature data sequence collected from the temperature sensor and the ambient humidity data sequence collected from the humidity sensor.

[0042] S202: When the voice pre-compensation processing module monitors voice data from an external microphone via the voice input interface, it intercepts the ambient temperature data subsequence collected within a previously preset time period from the ambient temperature data sequence, and intercepts the ambient humidity data subsequence collected within a previously preset time period from the ambient humidity data sequence.

[0043] Exemplarily, the preset time period can be, but is not limited to, 24 hours, 36 hours, 48 hours, etc., and is not limited herein.

[0044] S203: The voice pre-compensation processing module identifies the ambient temperature data subsequence and the ambient humidity data subsequence to determine whether they affect the voice playback of the speaker connected to the voice output interface.

[0045] Exemplarily, the specific implementation manners of S203 include, but are not limited to, the following two:

[0046] The first method: The voice pre-compensation processing module inputs the ambient temperature data subsequence and the ambient humidity data subsequence into a pre-trained voice playback impact determination model to determine whether there is an impact on the voice playback of the speaker connected to the voice output interface.

[0047] Among them, the voice playback impact determination model is obtained by training multiple third training samples input into a third neural network model. Each third training sample includes a historical ambient temperature data subsequence, a historical ambient humidity data subsequence, and a historical impact result label corresponding to whether there is an impact on the voice playback of the speaker.

[0048] The second method: The voice pre-compensation processing module determines the average ambient temperature within a previously preset duration according to the ambient temperature data subsequence, and determines the average ambient humidity within a previously preset duration according to the ambient humidity data subsequence; determines whether the average ambient temperature is greater than a set temperature threshold, and / or whether the average ambient humidity is greater than a set humidity threshold; if the average ambient temperature is greater than the set temperature threshold, and / or the average ambient humidity is greater than the set humidity threshold, it is determined that there is an impact on the voice playback of the speaker connected to the voice output interface.

[0049] S204: If there is an impact, the voice pre-compensation processing module inputs the ambient temperature data subsequence, the ambient temperature data subsequence, a preset simulation structure model of the speaker, and the material type for manufacturing the speaker into a pre-trained voice quality parameter determination model to obtain the short-term objective intelligibility (STOI) impact value, the perceptual objective voice quality assessment (POLQA) impact value, the segmented signal-to-noise ratio impact value, and the volume impact value for the voice played by the speaker.

[0050] Among them, the voice quality parameter determination model is obtained by training multiple first training samples input into a first neural network. Each first training sample includes a historical ambient temperature data subsequence, a historical ambient temperature data subsequence, a historical simulation structure model of the speaker, the material type for manufacturing the historical speaker, and the corresponding historical actual short-term objective intelligibility (STOI) impact value, historical actual perceptual objective voice quality assessment (POLQA) impact value, historical actual segmented signal-to-noise ratio impact value, and historical actual volume impact value.

[0051] Exemplarily, the short-term objective intelligibility (STOI) impact value can be understood as the difference between the STOI index of the speaker playing the voice normally when not affected and the STOI index of the speaker playing the voice when affected; the perceptual objective listening quality assessment (POLQA) impact value can be understood as the difference between the POLQA index of the speaker playing the voice normally when not affected and the POLQA index of the speaker playing the voice when affected; the segmental signal-to-noise ratio impact value can be understood as the difference between the segmental signal-to-noise ratio of the speaker playing the voice normally when not affected and the segmental signal-to-noise ratio of the speaker playing the voice when affected; the volume impact value can be understood as the difference between the volume of the speaker playing the voice normally when not affected and the volume of the STOI index of the speaker playing the voice when affected.

[0052] It should be noted that the short-term objective intelligibility (STOI) is an index for evaluating the quality of speech signals. Generally, the STOI value of speech signals is in the range of 0-1. The larger the STOI value, the higher the intelligibility (that is, the higher the quality of the speech signal). The perceptual objective listening quality assessment (POLQA) is also an index for evaluating the quality of speech signals. The speech signal can be rated based on the full-reference algorithm. It compares each sample of the reference speech signal (speaker side) with each corresponding sample of the degraded speech signal (listener side). The perceptual difference between the two speech signals is rated as the difference. After applying the masking function, the two speech signals are analyzed in the frequency domain (in critical bands). The unmasked difference between the two speech signal representations is counted as distortion. Finally, the accumulated distortion in the speech signal is mapped to the MOS test with quality levels from 1 to 5. Among them, the quality level is the POLQA index. The segmental signal-to-noise ratio refers to the ratio of the signal to noise in different frequency bands of the speech signal.

[0053] In addition, the method provided in the embodiments of the present application may further include:

[0054] Step 1: Determine whether there is a fault in the speaker according to the short-term objective intelligibility (STOI) impact value, the perceptual objective listening quality assessment (POLQA) impact value, the segmental signal-to-noise ratio impact value, and the volume impact value of the voice played by the speaker.

[0055] Step 1 may be specifically implemented as: According to the formula P = k 1 a + k 2 b + k 3 c + k 4d. Determine the comprehensive influence factor on the voice playback of the speaker. When the comprehensive influence factor on the voice playback of the speaker is greater than the set influence threshold, it is determined that the speaker has a fault.

[0056] Where P is the comprehensive influence factor on the voice playback of the speaker, a is the short-term objective intelligibility (STOI) influence value, b is the perceptual objective speech quality assessment (POLQA) influence value, c is the segmented signal-to-noise ratio influence value, d is the volume influence value, k 1 is the preset first weighting coefficient, k 2 is the preset second weighting coefficient, k 3 is the preset third weighting coefficient, k 4 is the preset fourth weighting coefficient, and k 1 + k 2 + k 3 + k 4 =1. For example, k 1 =k 2 =k 3 =k 4 =0.25.

[0057] Step 2: If there is a fault, output a prompt message indicating that the speaker has a fault to the display panel.

[0058] S205: Input the short-term objective intelligibility (STOI) influence value, the perceptual objective speech quality assessment (POLQA) influence value, the segmented signal-to-noise ratio influence value, and the volume influence value into the pre-trained voice adjustment parameter determination model to determine the gain adjustment parameter of the AGC module, the gain adjustment parameters of the DRC module in different time domains, and the gain adjustment parameters of the equalizer in different frequency domains.

[0059] Among them, the voice adjustment parameter determination model is obtained by training multiple second training samples input into the second neural network. Each second training sample includes the historical short-term objective intelligibility (STOI) influence value, the historical perceptual objective speech quality assessment (POLQA) influence value, the historical segmented signal-to-noise ratio influence value, the historical volume influence value, and the corresponding historical actual gain adjustment parameter of the AGC module, the historical actual gain adjustment parameters of the DRC module in different time domains, and the historical actual gain adjustment parameters of the equalizer in different frequency domains.

[0060] It should be noted that the automatic gain control (AGC) module refers to a module that can automatically adjust the gain of an amplifier circuit according to the signal strength. The AGC module is a closed-loop electronic circuit and a negative feedback system. It can be divided into a gain-controlled amplifier circuit and a control voltage formation circuit. The gain-controlled amplifier circuit is located in the forward amplification path, and its gain changes with the control voltage. The dynamic range control (DRC) module is generally used to dynamically adjust the amplitude of the audio output, suppress the volume within a certain range when the volume is large, and appropriately increase the volume when the volume is small, that is, used to control the audio output power to prevent the speaker from distorting and ensure clear sound even when playing at a low volume. The equalizer (EQ) is an electronic device that can adjust the amplification of electrical signals of various frequency components separately. By adjusting the electrical signals of various different frequencies, it compensates for the defects of the speaker and the sound field, and compensates and modifies various sound sources. The equalizer can adjust the electrical signals of the high-frequency, middle-frequency, and low-frequency bands separately.

[0061] S206: The voice pre-compensation processing module adjusts the gain control parameters configured in the AGC module according to the gain adjustment parameters of the AGC module, adjusts the gain control parameters configured in the DRC module according to the gain adjustment parameters of the DRC module in different time domains, and adjusts the gain control parameters configured in the equalizer according to the gain adjustment parameters of the equalizer in different frequency domains.

[0062] S207: The voice pre-compensation processing module controls the AGC module to process the voice data based on the adjusted gain control parameters of the AGC and output it to the DRC module, the DRC module to process the voice data based on the adjusted gain control parameters of the DRC module and output it to the equalizer, and the equalizer to process the voice data based on the adjusted gain control parameters of the equalizer and output it to the digital power amplifier module.

[0063] S208: The digital power amplifier module sequentially performs analog-to-digital conversion, filtering processing, amplification processing, and digital-to-analog conversion on the received voice data, and transmits it to the speaker for playback via the voice output interface.

[0064] In addition, the embodiment of the present application also provides a digital power amplifier device. The digital power amplifier device includes a housing, and the housing can be a rectangular box. The surface of the housing is provided with a voice input interface, a voice output interface, a temperature sensor, and a humidity sensor at intervals. Inside the housing, there are a voice pre-compensation processing module, an automatic gain control (AGC) module, a dynamic range control (DRC) module, an equalizer, and a digital power amplifier module. As Figure 1The voice input interface, AGC module, DRC module, equalizer, digital power amplifier module, and voice output interface are electrically connected in sequence. The voice input interface, temperature sensor, humidity sensor, AGC module, DRC module, and equalizer are respectively electrically connected to the voice pre-compensation processing module, and the voice pre-compensation processing module is used to execute the voice pre-compensation processing method in the corresponding embodiment of the present application. Figure 2 The voice pre-compensation processing method in the corresponding embodiment of the present application.

[0065] In addition, a power supply port is provided on the surface of the housing, and the power supply port is electrically connected to the voice pre-compensation processing module.

[0066] In addition, an embodiment of the present application further provides a storage medium storing a computer program, and when the computer program is executed by a processor, the computer is caused to execute the method executed by the voice pre-compensation processing module in the corresponding embodiment of the present application. Figure 2 The method executed by the voice pre-compensation processing module in the corresponding embodiment of the present application.

[0067] In addition, an embodiment of the present application further provides a computer program product including a computer program, and when the computer program is run, the computer is caused to execute the method executed by the voice pre-compensation processing module in the corresponding embodiment of the present application. Figure 2 The method executed by the voice pre-compensation processing module in the corresponding embodiment of the present application.

[0068] In summary, an embodiment of the present application provides a voice pre-compensation processing method and a digital power amplifier device. When the voice pre-compensation processing module determines that the environmental temperature data subsequence and the environmental humidity data subsequence affect the voice playback of the speaker connected to the voice output interface, the environmental temperature data subsequence, the environmental temperature data subsequence, the preset simulation structure model of the speaker, and the material type of the speaker are input into the pre-trained voice quality parameter determination model to obtain the short-term objective intelligibility (STOI) influence value, the perceptual objective speech quality assessment (POLQA) influence value, the segmental signal-to-noise ratio influence value, and the volume influence value on the voice played by the speaker. Since the voice quality parameter determination model is trained by inputting multiple first training samples into the first neural network, the accuracy of the output result is high.

[0069] The short-term objective intelligibility (STOI) influence value, the perceptual objective speech quality assessment (POLQA) influence value, the segmental signal-to-noise ratio influence value, and the volume influence value are input into the pre-trained voice adjustment parameter determination model to determine the gain adjustment parameter of the AGC module, the gain adjustment parameter of the DRC module in different time domains, the gain adjustment parameter of the equalizer in different frequency domains, and the gain adjustment parameter of the digital power amplifier module. Since the voice adjustment parameter determination model is trained by inputting multiple second training samples into the second neural network, the accuracy of the output result is high;

[0070] Furthermore, the voice pre-compensation processing module adjusts the gain control parameters configured in the AGC module according to the gain adjustment parameters of the AGC module, adjusts the gain control parameters configured in the DRC module according to the gain adjustment parameters of the DRC module in different time domains, and adjusts the gain control parameters configured in the equalizer according to the gain adjustment parameters of the equalizer in different frequency domains.

[0071] The voice pre-compensation processing module controls the AGC module to process voice data based on the adjusted gain control parameters of the AGC and output it to the DRC module, the DRC module processes voice data based on the adjusted gain control parameters of the DRC module and outputs it to the equalizer, and the equalizer processes voice data based on the adjusted gain control parameters of the equalizer and outputs it to the digital power amplifier module; the digital power amplifier module performs analog-to-digital conversion, filtering, amplification, and digital-to-analog conversion on the received voice data, and transmits it to the speaker for playback via the voice output interface. It can be understood that when the performance of the speaker is affected by the environment, since the played voice has been pre-compensated by the gain adjustment parameters of the AGC module, the gain adjustment parameters of the DRC module in different time domains, the gain adjustment parameters of the equalizer in different frequency domains, and the gain adjustment parameters of the digital power amplifier module, the STOI index, POLQA index, segmental signal-to-noise ratio, volume, and other indicators affecting voice quality of the voice can be compensated, so that the voice played by the speaker is no longer affected by the environmental erosion, and the voice quality is high, meeting the needs of users.

[0072] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A speech precompensation processing method, characterized in that: The invention is applied to a digital power amplifier device, the digital power amplifier device comprises a housing, a voice input interface, a voice output interface, a temperature sensor and a humidity sensor are arranged at intervals on the surface of the housing, a voice pre-compensation processing module, an automatic gain control AGC module, a dynamic range control DRC module, an equalizer and a digital power amplifier module are arranged in the housing, the voice input interface, the AGC module, the DRC module, the equalizer, the digital power amplifier module and the voice output interface are electrically connected in sequence, the voice input interface, the temperature sensor, the humidity sensor, the AGC module, the DRC module and the equalizer are electrically connected to the voice pre-compensation processing module respectively, and the method comprises: The speech pre-compensation processing module receives an ambient temperature data sequence collected by the temperature sensor and an ambient humidity data sequence collected by the humidity sensor; When the voice pre-compensation processing module detects voice data from an external microphone via the voice input interface, the voice pre-compensation processing module intercepts a subsequence of ambient temperature data collected within a previously preset time period from the ambient temperature data sequence, and intercepts a subsequence of ambient humidity data collected within a previously preset time period from the ambient humidity data sequence; The speech pre-compensation processing module identifies whether the ambient temperature data subsequence and the ambient humidity data subsequence have an impact on the speech playback of the speaker connected to the speech output interface; If there is an impact, the speech pre-compensation processing module inputs the ambient temperature data subsequence, the ambient temperature data subsequence, and the preset simulation structure model of the loudspeaker and the material type of the loudspeaker into the pre-trained speech quality parameter determination model to obtain a short-time objective and intelligible STOI impact value, a perceptual objective speech quality assessment POLQA impact value, a segmented signal-to-noise ratio impact value, and a volume impact value of the speech played by the loudspeaker, wherein the speech quality parameter determination model is obtained by inputting a plurality of first training samples into a first neural network for training, each of the first training samples including a historical ambient temperature data subsequence, a historical ambient temperature data subsequence, a simulation structure model of a historical loudspeaker, the material type of the historical loudspeaker, and a corresponding historical actual short-time objective and intelligible STOI impact value, a historical actual perceptual objective speech quality assessment POLQA impact value, a historical actual segmented signal-to-noise ratio impact value, and a historical actual volume impact value; Input the short-time objectively intelligible STOI impact value, the perceptual objective speech quality assessment POLQA impact value, the segmented signal-to-noise ratio impact value, and the volume impact value into a pre-trained speech adjustment parameter determination model to determine the gain adjustment parameter of the AGC module, the gain adjustment parameter of the DRC module in different time domains, and the gain adjustment parameter of the equalizer in different frequency domains, wherein the speech adjustment parameter determination model is obtained by inputting a plurality of second training samples into a second neural network for training, each of the second training samples including a historical short-time objectively intelligible STOI impact value, a historical perceptual objective speech quality assessment POLQA impact value, a historical segmented signal-to-noise ratio impact value, a historical volume impact value, and corresponding historical actual gain adjustment parameters of the AGC module, historical actual gain adjustment parameters of the DRC module in different time domains, and historical actual gain adjustment parameters of the equalizer in different frequency domains; The speech pre-compensation processing module adjusts the gain control parameters configured in the AGC module according to the gain adjustment parameters of the AGC module, adjusts the gain control parameters configured in the DRC module according to the gain adjustment parameters of the DRC module in different time domains, and adjusts the gain control parameters configured in the equalizer according to the gain adjustment parameters of the equalizer in different frequency domains; The speech pre-compensation processing module controls the AGC module to process the speech data based on the adjusted AGC gain control parameters and output the data to the DRC module; the DRC module processes the speech data based on the adjusted gain control parameters of the DRC module and outputs the data to the equalizer; the equalizer processes the speech data based on the adjusted gain control parameters of the equalizer and outputs the data to the digital power amplifier module; The digital power amplifier module sequentially performs analog-to-digital conversion, filtering, amplification, and digital-to-analog conversion on the received voice data, and transmits the data to the speaker for playback via the voice output interface.

2. The method according to claim 1, characterized in that The speech pre-compensation processing module identifies whether the ambient temperature data subsequence and the ambient humidity data subsequence have an impact on the speech playback of the speaker connected to the speech output interface, including: The speech pre-compensation processing module inputs the ambient temperature data subsequence and the ambient humidity data subsequence into a pre-trained speech playback impact determination model to determine whether there is an impact on the speech playback of the speaker connected to the speech output interface; wherein the speech playback impact determination model is obtained by inputting multiple third training samples into a third neural network model for training, and each of the third training samples includes a historical ambient temperature data subsequence, a historical ambient humidity data subsequence, and a corresponding historical impact result label of whether there is an impact on the speech playback of the speaker.

3. The method according to claim 1, characterized in that The speech pre-compensation processing module identifies whether the ambient temperature data subsequence and the ambient humidity data subsequence have an impact on the speech playback of the speaker connected to the speech output interface, including: The speech pre-compensation processing module determines the average value of the ambient temperature within the previous preset time period according to the ambient temperature data subsequence, and determines the average value of the ambient humidity within the previous preset time period according to the ambient humidity data subsequence; Determine whether the average value of the ambient temperature is greater than a set temperature threshold, and / or whether the average value of the ambient humidity is greater than a set humidity threshold; If the average value of the ambient temperature is greater than a set temperature threshold, and / or the average value of the ambient humidity is greater than a set humidity threshold, it is determined that there is an impact on the voice playback of the speaker connected to the voice output interface.

4. The method according to claim 1, characterized in that: The surface of the housing is also embedded with a display panel, and the display panel is electrically connected to the speech pre-compensation processing module. After obtaining the short-time objective and intelligible STOI impact value, the perceptual objective speech quality assessment POLQA impact value, the segmented signal-to-noise ratio impact value, and the volume impact value of the speech played by the speaker, the method further includes: Determine whether the speaker has a fault according to a short-time objectively intelligible STOI impact value, a perceptual objective speech quality assessment POLQA impact value, a segmented signal-to-noise ratio impact value, and a volume impact value of the speech played by the speaker; If there is a fault, prompt information indicating that there is a fault in the speaker is output to the display panel.

5. The method according to claim 4, characterized in that Determining whether the speaker has a fault according to a short-time objectively intelligible STOI impact value, a perceptual objective speech quality assessment POLQA impact value, a segmented signal-to-noise ratio impact value, and a volume impact value of the speech played by the speaker, including: According to the formula P=k1a+k2b+k3c+k4d, determine the comprehensive influence factor of the voice playback of the speaker, wherein P is the comprehensive influence factor of the voice playback of the speaker, a is the short-time objective and intelligible STOI influence value, b is the perceptual objective voice quality assessment POLQA influence value, c is the segmented signal-to-noise ratio influence value, d is the volume influence value, k1 is a preset first weighting coefficient, k2 is a preset second weighting coefficient, k3 is a preset third weighting coefficient, k4 is a preset fourth weighting coefficient, and k1+k2+k3+k4=1; When the comprehensive impact factor on the voice playback of the speaker is greater than a set impact threshold, it is determined that the speaker has a fault.

6. A digital power amplifier device, characterized in that: The digital power amplifier device includes a shell, and a voice input interface, a voice output interface, a temperature sensor and a humidity sensor are arranged at intervals on the surface of the shell. A voice pre-compensation processing module, an automatic gain control AGC module, a dynamic range control DRC module, an equalizer and a digital power amplifier module are arranged in the shell. The voice input interface, the AGC module, the DRC module, the equalizer, the digital power amplifier module and the voice output interface are electrically connected in sequence. The voice input interface, the temperature sensor, the humidity sensor, the AGC module, the DRC module and the equalizer are electrically connected to the voice pre-compensation processing module respectively. The voice pre-compensation processing module is used to execute any of the voice pre-compensation processing methods described in 1-5.

7. The digital power amplifier device according to claim 6, characterized in that: A power supply port is also provided on the surface of the shell, and the power supply port is electrically connected to the speech pre-compensation processing module.

8. The digital power amplifier device according to claim 6, characterized in that: The shell is a rectangular box body.

9. A storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the computer executes the method executed by the speech pre-compensation processing module as claimed in any one of claims 1 to 5.

10. A computer program product, comprising a computer program, which enables a computer to execute the method executed by the speech pre-compensation processing module as claimed in any one of claims 1 to 5 when the computer program is executed.

Images