A method and device for collecting voice signals using a microphone array

Abstract

A method and device for collecting voice signals for microphone arrays, the method comprising the steps of: obtaining multiple groups of voice signals, so that each group of voice signals includes voice signals from two alternate or adjacent microphones, converting and processing the two voice signals in each group respectively, obtaining digital signals corresponding to the two voice signals respectively, and obtaining a multi-channel stereo signal by synthesizing and processing the obtained digital signals. Since a cross-injection collection method is adopted for the microphone voice signals, the voice signals of adjacent or alternate microphones can be used as sampling references, which can not only enable the acquisition channel of each group of voice signals to have independent acquisition capabilities, but also realize the synchronous acquisition requirements of different acquisition positions, which is conducive to improving the mixed acquisition effect of voice signals, reducing the design difficulty of synchronous acquisition of multi-channel microphone arrays, and improving the expansibility of voice signal acquisition of microphone arrays.

Description

Technical Field

[0001] The present application relates to the field of speech signal processing, and in particular to a speech signal acquisition method and an acquisition device for a microphone array. Background Art

[0002] At present, microphone signal collectors generally use independent analog-to-digital conversion ADC to complete the digital acquisition of voice signals under the control of a microcontroller. Compared with the traditional independent analog signal ADC acquisition technology, dual-channel stereo acquisition technology has the advantages of low cost, low power consumption, high sampling bit number, programmable gain control, good synchronous acquisition characteristics, and fast AD conversion time. It is widely used in artificial intelligence voice signal acquisition terminals, digital communication cameras, and mobile device voice signal acquisition and processing fields, and is a key technology for realizing equipment digitization.

[0003] like Figure 1 As shown, the basic structure of the data acquisition using independent analog-to-digital conversion ADC in the prior art solution includes a microphone voice sensor unit microphone 1, a signal amplification unit 2, a digital-to-analog conversion ADC unit 3, a data read-write control unit 11, a clock generation unit 10, a PC host unit 12 and a storage unit 13. The clock generation unit 10 provides a clock signal with a frequency of fCLK, and the analog-to-digital conversion ADC circuit unit 3 realizes the process of converting the dual-channel stereo voice signal analog signal into a digital signal under the drive of the clock fCLK generated above, and then the data read-write control unit 11 uses the sampling clock of the same speed (frequency is fCLK) to read out and store the digital signal, and finally uses the storage unit 13 to store the digital signal, so as to obtain the desired voice signal.

[0004] for Figure 1 The illustrated technical solution separates the signal amplification circuit and the digital-to-analog conversion circuit. The amplification gain of the amplification circuit is difficult to be consistent, and the gain coefficient cannot be dynamically adjusted according to the amplitude of the input signal. Therefore, saturation distortion caused by over-discharge of the signal may occur. Please refer to Figure 2 The test results of the data collector of the prior art solution in the figure, wherein x1 shows two curves representing the waveforms of the collected speech signals of the two channels. In the experiment, the two microphones are placed at the same distance from the sound source. It can be seen from the curve results that at the characteristic points of the signal, such as the peaks and troughs, the time points at which they appear on the time axis are at the same moment, and the synchronization characteristics of the two signal acquisitions can be qualitatively judged. If the accuracy of the synchronized acquisition needs to be quantified, the cross-correlation function can be used for quantitative evaluation.

[0005] It can be seen from this that the existing technology has the following shortcomings: (1) The sampling rate of the voice signal collector is limited by the number of collection channels and the main frequency of the core controller of the data read-write control unit and the digital transmission bandwidth. For example, when the PC host computer communicates with the data read-write control unit through a USB cable, due to the limitation of the USB data transmission bandwidth, the number of collector channels with a sampling rate of 200K and a quantization bit of 12 bits when reading data from the USB interface cannot be expanded indefinitely, and usually can only accommodate about 10 channels of data for simultaneous collection.

[0006] (2) Due to its high clock frequency, the ADC of the digital-to-analog converter will encounter the problem of insufficient clock driving capability and signal attenuation leading to asynchronous acquisition when designing the multi-channel microphone voice signal acquisition circuit, which makes the circuit design complex and costly. In addition, the preamplifier circuit of the ordinary collector will have a large difference in the amplification factor of each channel due to the differences in the parameters of electronic devices such as resistors and capacitors during the circuit design and integration process, and is easily disturbed by changes in the external environment, resulting in the asynchrony of the analog-to-digital converter signal acquisition and severe noise. Summary of the invention

[0007] The technical problem actually solved by this application is how to collect voice signals of a microphone array through multiple channels and obtain a better collection effect. To solve the above problem, this application provides a voice signal collection method and a collection device for a microphone array.

[0008] According to a first aspect, an embodiment of the present application provides a method for collecting speech signals using a microphone array, comprising the following steps:

[0009] Acquire multiple groups of speech signals, each group of speech signals includes speech signals from two alternate or adjacent microphones;

[0010] Converting the two voice signals in each group respectively to obtain digital signals corresponding to the two voice signals;

[0011] A multi-channel stereo signal is obtained according to the digital signal synthesis process.

[0012] The obtaining of multiple groups of voice signals comprises:

[0013] Acquire a voice signal from a microphone, and acquire another voice signal from another microphone spaced apart from the microphone, so that a group of voice signals includes the voice signals of the two spaced apart microphones; and / or,

[0014] A voice signal is acquired from one microphone, and another voice signal is acquired from another microphone adjacent to the microphone, so that a group of voice signals includes the voice signals of two adjacently arranged microphones.

[0015] The two voice signals in each group are converted and processed respectively to obtain digital signals corresponding to the two voice signals respectively, including:

[0016] For the two voice signals in each group, synchronous analog-to-digital conversion is performed on the two voice signals according to a preset sampling bit number and a preset sampling rate to obtain digital signals corresponding to the two voice signals respectively.

[0017] Before converting the two voice signals in each group, a signal amplification step is also included, and the signal amplification step includes:

[0018] The voice signal of each channel is amplified according to the preset amplification gain, so as to perform analog-to-digital conversion on the amplified voice signal.

[0019] The step of obtaining a multi-channel stereo signal according to the digital signal synthesis process comprises:

[0020] The digital signals after each group of voice signal conversion and processing are stored separately;

[0021] The multi-channel stereo signal is obtained by synthesizing and processing the synchronously stored digital signals.

[0022] According to a second aspect, an embodiment of the present application provides a speech signal collection device for a microphone array, comprising:

[0023] A plurality of acquisition modules, used for respectively acquiring a group of speech signals, each group of speech signals including speech signals of two alternate or adjacent microphones;

[0024] A plurality of conversion modules are connected to the plurality of acquisition modules in a one-to-one correspondence, and are used to convert two voice signals in a group of voice signals obtained by each acquisition module to obtain digital signals corresponding to the two voice signals respectively;

[0025] The processing module is connected to the plurality of acquisition modules and is used to obtain a multi-channel stereo signal according to the synthesis processing of the digital signals.

[0026] The acquisition module acquires one voice signal from one microphone, and acquires another voice signal from another microphone spaced apart from the microphone, so that a group of acquired voice signals includes voice signals from two spaced apart microphones; and / or,

[0027] The acquisition module acquires one voice signal from one microphone, and acquires another voice signal from another microphone adjacent to the microphone, so that a group of acquired voice signals includes voice signals from two adjacently arranged microphones.

[0028] Each conversion module includes two analog-to-digital conversion circuits, a data read-write control unit, and a clock generation unit, wherein the two analog-to-digital conversion circuits are respectively connected to the acquisition module connected to the conversion module, and the data read-write control unit and the clock generation unit are both connected to the two analog-to-digital conversion circuits;

[0029] For two voice signals in a group of voice signals obtained by each acquisition module, the data read and write control unit and the clock generation unit control the two analog-to-digital conversion circuits to perform synchronous analog-to-digital conversion on the two voice signals according to a preset number of sampling bits and a preset sampling rate to obtain digital signals corresponding to the two voice signals respectively.

[0030] The speech signal acquisition device further comprises a plurality of signal amplification modules, each of which is arranged between each acquisition module and a conversion module connected to the acquisition module;

[0031] The plurality of signal amplifying modules amplify the voice signals of each channel respectively according to a preset amplification gain, so as to perform analog-to-digital conversion on the amplified voice signals.

[0032] The speech signal collection device further comprises a plurality of storage units, wherein the plurality of storage units are respectively connected to the plurality of conversion modules;

[0033] The plurality of storage units respectively store the digital signals after each group of speech signal conversion processing, so that the processing module obtains the synchronously stored digital signals from the plurality of storage units and synthesizes and processes them to obtain the multi-channel stereo signal.

[0034] The beneficial effects of the storage unit voice signal acquisition method of the present application are:

[0035] According to the above-mentioned embodiment, a method and device for collecting voice signals for microphone arrays are provided. The method comprises the following steps: obtaining multiple groups of voice signals, so that each group of voice signals includes voice signals of two alternate or adjacent microphones, converting and processing the two voice signals in each group respectively, obtaining digital signals corresponding to the two voice signals respectively, and synthesizing and processing the obtained digital signals to obtain a multi-channel stereo signal. In the first aspect, a cross-injection collection method is adopted for the microphone voice signals, so that the voice signals of adjacent or alternate microphones can be used as sampling references, which can not only enable the acquisition channel of each group of voice signals to have independent acquisition capabilities, but also realize the synchronous acquisition requirements of different acquisition positions, which is conducive to improving the mixed acquisition effect of voice signals, reducing the design difficulty of synchronous acquisition of multi-channel microphone arrays, and improving the scalability of voice signal acquisition of microphone arrays; in the second aspect, since a symmetrical amplification and conversion structure is used to synchronously amplify and convert a group of voice signals, it is conducive to ensuring the consistency and synchronization of the amplification and conversion process, and avoiding noise and loss caused by asynchrony as much as possible. true situation; thirdly, the acquisition device constructed based on the voice signal acquisition method adopts a data read and write control module to control the conversion process of the voice signal, so that the converted digital signal can be stored and transmitted through a double buffering mechanism; fourthly, since the voice signal acquisition device adopts multiple acquisition modules, multiple conversion modules and multiple storage units to jointly constitute the acquisition channels of multiple voice signals, it not only reduces the design difficulty of synchronous acquisition of voice signals of large microphone arrays, but also greatly improves the bandwidth of signal acquisition, and also makes the wiring structure of the microphone array better optimized, which is beneficial to improve the signal-to-noise ratio of the signal and avoids the situation of inconsistent signal routing length and difficult maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a structural schematic diagram of a speech signal collector of an existing microphone array;

[0037] Figure 2 This is a test effect diagram of the voice signal collector of the existing microphone array;

[0038] Figure 3 A simplified structural diagram of a speech signal acquisition device for a one-dimensional microphone array according to an embodiment;

[0039] Figure 4 A detailed structural diagram of a speech signal acquisition device for a one-dimensional microphone array according to an embodiment;

[0040] Figure 5 A simplified structural diagram of a speech signal acquisition device for a two-dimensional microphone array according to an embodiment;

[0041] Figure 6is a flow chart of a method for collecting speech signals using a microphone array;

[0042] Figure 7 A schematic diagram of data flow direction of a method for collecting voice signals using a microphone array according to an embodiment of the present invention;

[0043] Figure 8 This is one of the test effect diagrams of the voice signal acquisition device of the microphone array of the present application technology;

[0044] Fig. 9 This is the second diagram of the synchronous test effect of the voice signal acquisition device of the microphone array of the present application technology. DETAILED DESCRIPTION

[0045] The present application is further described in detail below through specific implementation methods in conjunction with the accompanying drawings.

[0046] Embodiment 1:

[0047] Please refer to Figure 3 , a speech signal acquisition device for a microphone array, which includes multiple acquisition modules (such as 111, 112, 113, 114), multiple conversion modules (such as 121, 122, 123, 124) and a processing module 13, which are described below respectively.

[0048] It should be noted that the microphone array of the present application is composed of multiple microphones arranged in rows and columns to form a one-dimensional array or a two-dimensional array, or even a multi-dimensional array. In this embodiment, a one-dimensional column array formed by multiple microphones (such as M1, M2, M3, M4, M5) is used as an example for explanation. The microphone here is used to receive the original voice signal and convert the natural voice signal into an analog signal output. The microphone can be a MEMS microphone sensor with high sensitivity, or an electret or capacitive microphone sensor. The specific type and specific number of the microphone are not limited here.

[0049] A plurality of acquisition modules (111, 112, 113, 114) are used to respectively acquire a group of voice signals, each group of voice signals including voice signals of two alternate or adjacent microphones. In one embodiment, the acquisition module acquires one voice signal from one microphone, and acquires another voice signal from another microphone spaced apart from the microphone, so that the acquired group of voice signals includes voice signals of two alternately spaced microphones, as specifically shown in the acquisition module. Of course, the voice signals of the microphones can also be acquired in another way, and the acquisition module acquires one voice signal from one microphone, and acquires another voice signal from another microphone adjacent to the microphone, so that the acquired group of voice signals includes voice signals of two adjacently spaced microphones, as specifically shown in the acquisition module 111. In this embodiment, it is preferred to adopt the acquisition method of two types of microphone voice signals, namely, the alternate and adjacent ones. The voice signal acquisition method of the adjacent microphones can be adopted for the edge microphones, and the voice signal acquisition mode of the alternate microphones can be adopted for the middle microphones.

[0050] It should be noted that the acquisition modules (111, 112, 113, 114) can be ports of communication lines. For example, the acquisition module 111 is used as a communication port to acquire voice signals from the microphone M1 and the microphone M2 respectively. Figure 4 The one-dimensional microphone array shown in the figure mainly includes 1, 2, ...m...n microphones, and the acquisition modules (111, 112, ..., 114) are used to collect voice signals from these microphones, wherein m and n are only the arrangement numbers of the microphones, and multiple identical modules are also omitted between the acquisition modules 112 to 114. Therefore, this should not be understood as a limitation on the number of microphones and the number of acquisition modules (as well as the number of conversion modules).

[0051] A plurality of conversion modules (121, 122, 123, 124) are connected to a plurality of acquisition modules (111, 112, 113, 114) in a one-to-one correspondence, and are used to convert two voice signals in a group of voice signals obtained by each acquisition module to obtain digital signals corresponding to the two voice signals; when the conversion modules (121, 122, 123, 124) perform analog-to-digital conversion, a unified sampling signal can be provided to each conversion module to control the analog-to-digital conversion process. In one embodiment, see Figure 4For each conversion module, it includes two analog-to-digital conversion circuits, a data read-write control unit and a clock generating unit. The two analog-to-digital conversion circuits are respectively connected to the acquisition modules connected to the conversion module, and the data read-write control unit and the clock generating unit are both connected to the two analog-to-digital conversion circuits; for two voice signals in a group of voice signals obtained by each acquisition module, the data read-write control unit and the clock generating unit control the two analog-to-digital conversion circuits to perform synchronous analog-to-digital conversion on the two voice signals according to a preset number of sampling bits and a preset sampling rate, so as to obtain digital signals corresponding to the two voice signals respectively.

[0052] In one specific embodiment, see Figure 4 For example, the conversion module 121 includes two analog-to-digital conversion circuits (1211, 1212), a data read-write control unit 1213 and a clock generation unit 1214, wherein the analog-to-digital conversion circuits (1211, 1212) are common analog-to-digital conversion chips, such as WM8978 (the chip WM8978 can provide a standard audio DAC clock configuration, 256fs MCLK for DAC and ADC), which has the characteristics of adjustable sampling bits and sampling rate, and the data read-write control unit 1213 can control the sampling bits of the analog-to-digital conversion circuits (1211, 1212) to read out the digital signal obtained by the analog-to-digital conversion; the data read-write control unit 1213 The core processor of the control unit 1213 has an IIS interface dedicated to reading and writing digital voice signals, and the interface includes a clock signal interface for data reading, an analog-to-digital conversion data reading interface, and a digital audio output interface; the clock generation unit 1214 is a common clock chip, which can provide a frequency signal for the conversion process for the analog-to-digital conversion circuit (1211, 1212), and control the sampling rate of the analog-to-digital conversion through the frequency signal; usually, the analog-to-digital conversion circuit 1211 and the analog-to-digital conversion circuit 1212 will cache the conversion results in two data buffers in turn, and the core processor of the data read and write control unit 1213 will automatically read data from these two data buffers.

[0053] It should be noted that the sampling bit number of the analog-to-digital conversion process can be selected from 8 bits, 16 bits, and 24 bits, and the sampling rate can be selected from 8Khz, 11.025Khz, 16Khz, 22.05Khz, 32Khz, 44.1Khz, 48Khz, 88.2Khz, 96Khz, 176.4Khz, and 192Khz, and programmed. In this embodiment, in order to ensure the consistency and synchronization of sampling, each conversion module can have the same sampling bit number and sampling rate.

[0054] It should be noted that STM32F429 can be used as the core processor to complete data reading, writing and control operations. Then, the data reading and writing and control unit 1213 uses a ping-pong operation to read out the voice data at high speed, so that the analog-to-digital conversion circuit (1211, 1213) has a higher sampling rate and stability. It should be noted that since the analog-to-digital conversion circuit 1211 and the analog-to-digital conversion circuit 1212 use the same clock, the two-channel analog-to-digital conversion inside the conversion unit can be performed in parallel and synchronously. In order to ensure the stability of the clock signal, the clock generation unit 1214 preferably uses a high-precision phase-locked loop circuit.

[0055] It should be noted that the multi-bit oversampling ADC technology that can be used inside the analog-to-digital conversion circuit (1211, 1212) of this embodiment can filter out the common-mode noise mixed in the signal through multi-bit feedback and high oversampling rate to reduce the influence of pulse jitter and high-frequency noise. For example, an optional high-pass filter and an adjustable notch filter are provided, which have a variable center frequency and bandwidth, and the center frequency and bandwidth are adjusted by two coefficients a0 and a1, and a0 and a1 are set by registers NFA0[13:0] and NFA1[13:0] respectively. So that a0=(1-tan(wb / 2) / (1+tan(wb / 2))),a1=-(1+ a0)cos(w0), where w0=2πfc / fs,wb=2πfb / fs,fc is the center frequency, fb is the -3dB bandwidth, and fs is the sampling frequency; and the actual value of register NFA0[13:0] is -a0 multiplied by 2 to the 13th power, and the actual value of register NFA1[13:0] is -a1 multiplied by 2 to the 12th power.

[0056] It should be noted that the structures and functions of the other conversion modules (122, 124) are the same as those of the conversion module 121. Therefore, the conversion modules (122, 124) can be understood with reference to the conversion module 121, and will not be described in detail here.

[0057] Further, see Figure 4 The speech signal acquisition device further comprises a plurality of signal amplification modules (141, 142, ..., 144), each of which is arranged between each acquisition module and a conversion module connected to the acquisition module. The signal amplification modules (141, 142, ..., 144) amplify each voice signal according to a preset amplification gain, so as to perform analog-to-digital conversion on the amplified voice signal. In a specific embodiment, see Figure 4The signal amplification module 141 includes a signal amplification circuit 1411 and a signal amplification circuit 1412, which respectively amplify the two voice signals acquired by the acquisition module 111, and transmit the amplified voice signals to the analog-to-digital conversion circuit 1211 and the analog-to-digital conversion circuit 1212 respectively.

[0058] It should be noted that the signal amplification module (141, 142, ..., 144) can use a common signal amplification chip, wherein the gain of the signal amplification circuit can be set by programming, and the gain of the signal amplification circuit can be adjusted in the range of -12dB to +35.25dB, and the adjustment step can be set to 0.75dB; specifically, the gain coefficient can be set through the IIC interface of the STM32F429. In this embodiment, in order to ensure the consistency of voice signal amplification, the signal amplification modules (141, 142, ..., 144) are preferably set to the same gain coefficient.

[0059] Those skilled in the art will appreciate that the clock frequencies of the various clock generating units may differ slightly, and the gain coefficients of the various signal generating circuits cannot be absolutely consistent. Therefore, the present invention adopts a cross-injection voice signal acquisition method to sample the voice signals in groups. In this way, the inconsistency of the voice signals in the analog-to-digital conversion process can be avoided to the greatest extent.

[0060] Furthermore, the speech signal acquisition device further includes a plurality of storage units (151, 152, ..., 154), which are respectively connected to the plurality of conversion modules (121, 122, ..., 124). In one embodiment, the plurality of storage units (151, 152, ..., 154) respectively store the digital signals after each group of speech signal conversion processing, so that the processing module 13 obtains the synchronously stored digital signals from the storage units (151, 152, ..., 154) and synthesizes and processes to obtain the multi-channel stereo signal.

[0061] It should be noted that the storage unit (151, 152, ..., 154) can use a TF card for data storage. When the number of microphones in the microphone array increases, it becomes difficult to read all digital signals from the data read-write control unit in real time due to the limitation of the data transmission bandwidth. The present invention first caches the data in the TF card. When a burst of data collection is completed, on the one hand, it is read from the TF card, and on the other hand, the data in the TF card can be read in sequence by the data read-write control unit and uploaded to the processing module 13 through a USB port, so that the processing module can obtain the desired digital voice signal.

[0062] For reference Figure 3 and Figure 4The processing module 13 is connected to a plurality of acquisition modules (121, 122, 123, 124) and is used for synthesizing and processing the read digital signal (ie, digital voice signal) to obtain a multi-channel stereo signal.

[0063] It should be noted that here, a group of digital signals or multiple groups of digital signals can be superimposed and mixed to achieve the purpose of synthesizing and processing these digital signals, so that each synthesized audio signal forms a channel, so that multiple channels can be formed according to multiple synthesized audio signals, and finally a multi-channel stereo signal is obtained. These stereo signals are amplified and played respectively, and discernible stereo sound can be produced. When people directly hear stereo sound in a three-dimensional space, in addition to being able to feel the loudness, pitch and timbre of the sound, they can also feel their direction and level.

[0064] In another embodiment, it can be seen Figure 5 , the speech signal acquisition device disclosed in the present application is used to acquire speech signals from a two-dimensional cross microphone array, and the cross microphone array includes not only microphones M1-M5 in vertical columns, but also microphones in horizontal columns (such as M6, M7, M3, M8, and M9). Multiple acquisition modules and multiple conversion modules are respectively connected to the microphones in intervals and adjacently, so that each acquisition module has a cross-injection feature of two speech signals. Moreover, a unified sampling signal can be set for the clock generation unit in each conversion module, so that each conversion module has a consistent sampling rate. Of course, the microphone array can also have other composition methods, but they can all be connected using the speech signal acquisition device disclosed in the present application, and they will not be explained one by one here.

[0065] Embodiment 2 of the voice signal acquisition method:

[0066] Please refer to Figure 6 Based on the voice signal acquisition device claimed for protection in Example 1, the present application also discloses a voice signal acquisition method for a microphone array, which mainly includes steps S210-S230, which are described below respectively.

[0067] Step S210: acquiring multiple groups of voice signals, each group of voice signals including voice signals from two alternate or adjacent microphones.

[0068] In one embodiment, for example Figure 2, the acquisition module (111, 112, 113, 114) acquires one voice signal from a microphone, and acquires another voice signal from another microphone spaced apart from the microphone, so that a group of voice signals includes the voice signals of the two microphones spaced apart; and / or, the acquisition module (111, 112, 113, 114) acquires one voice signal from a microphone, and acquires another voice signal from another microphone adjacent to the microphone, so that a group of voice signals includes the voice signals of the two microphones adjacent to each other.

[0069] Step S220: convert the two voice signals in each group respectively to obtain digital signals corresponding to the two voice signals.

[0070] In one embodiment, for two voice signals in each group, synchronous analog-to-digital conversion is performed on the two voice signals according to a preset number of sampling bits and a preset sampling rate to obtain digital signals corresponding to the two voice signals. Figure 2 The data read-write control unit 1213 and the clock generation unit 1214 control the two analog-to-digital conversion circuits (1211, 1212) to perform synchronous analog-to-digital conversion on the two voice signals according to a preset number of sampling bits and a preset sampling rate to obtain digital signals corresponding to the two voice signals respectively.

[0071] It should be noted that the sampling bit number of the analog-to-digital conversion process can be selected from 8 bits, 16 bits, and 24 bits, and the sampling rate can be selected from 8Khz, 11.025Khz, 16Khz, 22.05Khz, 32Khz, 44.1Khz, 48Khz, 88.2Khz, 96Khz, 176.4Khz, and 192Khz, and programmed. In this embodiment, in order to ensure the consistency and synchronization of sampling, each conversion module can have the same sampling bit number and sampling rate.

[0072] In another embodiment, after step S210 and before step S220, an amplification step may be further included, the amplification step including: amplifying each voice signal according to a preset amplification gain, so as to perform analog-to-digital conversion on the amplified voice signal. Figure 2 The signal amplification module 141 includes a signal amplification circuit 1411 and a signal amplification circuit 1412, which respectively amplify the two voice signals acquired by the acquisition module 111, and transmit the amplified voice signals to the analog-to-digital conversion circuit 1211 and the analog-to-digital conversion circuit 1212 respectively.

[0073] It should be noted that the signal amplification module can use a common signal amplification chip, wherein the gain of the signal amplification circuit can be set by programming, and the gain of the signal amplification circuit can be adjusted in the range of -12dB to +35.25dB, and the adjustment step can be set to 0.75dB. In this embodiment, in order to ensure the consistency of voice signal amplification, the signal amplification modules (141, 142, ..., 144) are preferably set to the same gain coefficient.

[0074] Step S230: synthesizing and processing the digital signal obtained in step S220 to obtain a multi-channel stereo signal.

[0075] In one embodiment, step S230 specifically includes: storing the digital signals after each group of voice signal conversion processing respectively, and then synthesizing and processing the synchronously stored digital signals to obtain a multi-channel stereo signal. Figure 2 , multiple storage units (151, 152, ..., 154) respectively store the digital signals after each group of voice signal conversion processing, so that the processing module 13 obtains the synchronously stored digital signals from the storage units (151, 152, ..., 154) and synthesizes them to obtain the multi-channel stereo signal.

[0076] It should be noted that here, a group of digital signals or multiple groups of digital signals can be superimposed and mixed to achieve the purpose of synthesizing and processing these digital signals, so that each synthesized audio signal forms a channel, so that multiple channels can be formed according to multiple synthesized audio signals, and finally a multi-channel stereo signal is obtained. These stereo signals are amplified and played respectively, and discernible stereo sound can be produced. When people directly hear stereo sound in a three-dimensional space, in addition to being able to feel the loudness, pitch and timbre of the sound, they can also feel their direction and level.

[0077] In order to enable those skilled in the art to clearly and accurately understand the technical solution of the present application, the speech signal acquisition method is described in detail herein. Figure 7-Figure 9 The technical principles and technical effects of the technical solution are explained in detail.

[0078] Please refer to Figure 7In order to realize signal acquisition without delaying the system's signal processing and to improve the real-time performance of signal acquisition and processing, the present invention adopts a double buffer FOFO structure. The data stream after AD conversion is first stored in M0AR (memory 1) through DMA. When memory 1 is full, the hardware will automatically switch the buffer for receiving data to the next buffer M1AR (memory 2). At the same time, the program writes the data in M0AR (memory 1) to the SD card through DMA and saves it as a wav file. When M1AR (memory 2) is full, it automatically switches to M0AR (memory 1). In summary, whenever a buffer is full, the pointer of the buffer will spontaneously switch to the next buffer and store the data, thereby realizing uninterrupted data acquisition and storage of wav files.

[0079] Please refer to Figure 8 , is the voice signal waveform collected by the existing general data collector. In the figure, x1 is the signal waveform collected by the first channel of the existing dual-channel synchronous collector from a target microphone sensor, and x2 is the signal waveform collected by the second channel of the existing dual-channel synchronous collector from the same microphone connected to the first channel. Since the signal amplification part of the voice signal input end of the existing general data collector does not have an automatic gain control circuit in hardware, the amplification circuit fails to adjust the gain coefficient according to the amplitude of the input signal. When the amplitude of the external voice signal is large or when the microphone is close to the sound source, the acquisition is prone to excessive signal amplification, resulting in saturation distortion.

[0080] Please refer to Fig. 9 In order to verify the synchronization characteristics of the cross-injection structure for voice signal acquisition provided by the technical solution of this application, this application adopts the cross-correlation operation of the two digital voice signals collected to determine whether the coordinate position of the maximum point of the cross-correlation operation function is at the zero point and the left-right symmetry index of the cross-correlation function as the basis for evaluating the synchronous acquisition characteristics of the collector and the symmetry of the hardware design. The peak point of the function curve after the cross-correlation operation of the two collected signals is at the zero point, so the voice signals collected by the two data acquisition channels have good synchronization.

[0081] To sum up, the technical solution provided by the present application can reduce the difficulty of signal acquisition of the microphone array, improve the synchronization of the signal acquisition process of the microphone array, thereby reducing the design difficulty of the voice signal acquisition device, so that the design of the voice signal acquisition device is no longer subject to the bottleneck of signal acquisition synchronization, and the signal-to-noise ratio of the voice signal collected by the voice signal acquisition device is greatly improved, and the data transmission and data storage break through the bandwidth limitations of the existing technology. In addition, the voice signal collector adopts an orderly data caching mechanism inside the data reading and writing control unit, and has also been well optimized in terms of electronic circuit wiring.

[0082] The above contents are further detailed descriptions of the present application in combination with specific implementation methods, and it cannot be determined that the specific implementation of the present application is limited to these descriptions. For ordinary technicians in the technical field to which the present application belongs, several simple deductions or substitutions can be made without departing from the inventive concept of the present application.

Claims

1. A method for collecting speech signals using a microphone array, characterized in that: The following steps are involved: Acquire multiple groups of speech signals, each group of speech signals includes speech signals from two alternate or adjacent microphones; Converting the two voice signals in each group respectively to obtain digital signals corresponding to the two voice signals; Obtaining a multi-channel stereo signal according to the digital signal synthesis process; Wherein, obtaining multiple groups of voice signals includes: obtaining one voice signal from a microphone, and obtaining another voice signal from another microphone spaced apart from the microphone, so that one group of voice signals includes the voice signals of the two microphones spaced apart; and / or obtaining one voice signal from a microphone, and obtaining another voice signal from another microphone adjacent to the microphone, so that one group of voice signals includes the voice signals of the two microphones spaced apart; The two-way voice signals in each group are converted and processed respectively to obtain digital signals corresponding to the two-way voice signals respectively, including: for the two-way voice signals in each group, synchronous analog-to-digital conversion is performed on the two-way voice signals according to a preset number of sampling bits and a preset sampling rate to obtain digital signals corresponding to the two-way voice signals respectively; The voice signals in the same group are configured to be driven by the same clock generator for analog-to-digital conversion, and the voice signals in different groups are configured to be driven by different clock generators for analog-to-digital conversion; Among them, obtaining a multi-channel stereo signal according to the digital signal synthesis processing includes: storing the digital signals after each group of voice signal conversion processing separately; superimposing and mixing the obtained one or more groups of digital signals to achieve digital signal synthesis processing, so that each synthesized audio signal forms a channel; forming multiple channels according to multiple synthesized audio signals to obtain a multi-channel stereo signal.

2. The method for collecting speech signals according to claim 1, wherein: Before converting the two voice signals in each group, a signal amplification step is also included, and the signal amplification step includes: The voice signal of each channel is amplified according to the preset amplification gain, so as to perform analog-to-digital conversion on the amplified voice signal.

3. A speech signal acquisition device for a microphone array, characterized in that: include: A plurality of acquisition modules, used for respectively acquiring a group of speech signals, each group of speech signals including speech signals of two alternate or adjacent microphones; A plurality of conversion modules are connected to the plurality of acquisition modules in a one-to-one correspondence, and are used to convert two voice signals in a group of voice signals obtained by each acquisition module to obtain digital signals corresponding to the two voice signals respectively; A processing module, connected to the plurality of acquisition modules, for obtaining a multi-channel stereo signal according to the synthesis processing of the digital signal; The acquisition module acquires one voice signal from one microphone, and acquires another voice signal from another microphone spaced apart from the microphone, so that a group of voice signals acquired includes voice signals from two microphones spaced apart from each other; and / or, the acquisition module acquires one voice signal from one microphone, and acquires another voice signal from another microphone adjacent to the microphone, so that a group of voice signals acquired includes voice signals from two microphones spaced apart from each other; For each conversion module, two analog-to-digital conversion circuits, a data read-write control unit and a clock generating unit are included, and the two analog-to-digital conversion circuits are respectively connected to the acquisition module connected to the conversion module, and the data read-write control unit and the clock generating unit are both connected to the two analog-to-digital conversion circuits; for two voice signals in a group of voice signals obtained by each acquisition module, the data read-write control unit and the clock generating unit control the two analog-to-digital conversion circuits to perform synchronous analog-to-digital conversion on the two voice signals according to a preset number of sampling bits and a preset sampling rate, so as to obtain digital signals corresponding to the two voice signals respectively; the voice signals in the same group are configured to be driven by the same clock generator for analog-to-digital conversion, and the voice signals in different groups are configured to be driven by different clock generators for analog-to-digital conversion; A plurality of storage units are respectively connected to the plurality of conversion modules; the plurality of storage units respectively store the digital signals after each group of voice signal conversion processing, the processing module superimposes and mixes the obtained one or more groups of digital signals to achieve digital signal synthesis processing, so that each synthesized audio signal forms a channel; multiple channels are formed according to the multiple synthesized audio signals to obtain a multi-channel stereo signal.

4. The voice signal collection device according to claim 3, characterized in that: It also includes multiple signal amplification modules, each of which is arranged between each acquisition module and a conversion module connected to the acquisition module; the multiple signal amplification modules amplify each voice signal according to a preset amplification gain to perform analog-to-digital conversion on the amplified voice signal.

Images