An echo cancellation and environmental noise reduction method and device based on double microphone cooperation

Abstract

This invention discloses a method and apparatus for echo cancellation and environmental noise reduction based on dual-microphone collaboration, belonging to the field of speech signal processing technology. The method includes: providing a backtracking microphone and a speech microphone, both of which are enclosed; placing the backtracking microphone at a distance of 5mm or less from a speaker to collect the speaker's audio signal as a reference signal for echo cancellation, and collecting a first ambient sound; placing the speech microphone at a distance of 10cm or more from the speaker to collect the user's speech signal and a second ambient sound; canceling the speaker echo in the user's speech signal based on the reference signal; performing a differential operation on the first and second ambient sounds to eliminate environmental noise and output a clean user speech signal. This invention has the advantages of accurate echo cancellation, efficient environmental noise reduction, strong anti-interference ability, simple structure, and controllable cost.

Description

Technical Field

[0001] This invention relates to the field of speech signal processing technology, and in particular to a method and apparatus for echo cancellation and environmental noise reduction based on dual-microphone collaboration. Background Technology

[0002] Currently, echo cancellation and ambient noise reduction technologies for voice acquisition devices are widely used in electronic devices such as smart speakers, headphones, and conference phones. Existing technologies in this field can be mainly divided into the following two categories.

[0003] I. Single-microphone noise reduction solution This solution is commonly used in low-cost, low-end devices, capturing speech using only a single microphone and relying on software algorithms for noise reduction. Its advantages lie in its simple structure and low cost, but it cannot effectively distinguish between echoes, human voices, and environmental noise, resulting in residual echoes, difficulty in eliminating background noise, and easy distortion of the speech signal. This makes it difficult to meet the speech clarity requirements of mid-to-high-end devices.

[0004] II. Dual-microphone and multi-microphone solutions This is the current mainstream approach, using multiple microphones to collect signals for noise reduction. However, dual-microphone solutions often lack clear layout constraints, with microphones placed randomly, making it difficult to effectively distinguish between speaker echoes and human voice signals, resulting in incomplete echo cancellation. While multi-microphone solutions offer relatively better noise reduction, they are structurally complex, costly, and most lack anti-interference measures (such as not isolating the microphones). Furthermore, in some dual-microphone solutions, the two microphones do not coordinate effectively, leading to lower noise reduction efficiency.

[0005] In summary, existing technologies all suffer from problems such as inaccurate echo cancellation, limited noise reduction effect, weak anti-interference ability, complex structure, or poor adaptability. Summary of the Invention

[0006] The purpose of this invention is to provide a method and apparatus for echo cancellation and environmental noise reduction based on dual-microphone collaboration. By differentiating and enclosing the echo microphone and the voice microphone, the invention solves the problems of inaccurate echo cancellation, poor environmental noise reduction effect, weak anti-interference ability and complex structure in the prior art.

[0007] To achieve the above objectives, in a first aspect, the present invention provides a method for echo cancellation and environmental noise reduction based on dual-microphone collaboration, comprising the following steps: S1. Provide a sampling microphone and a voice microphone, and enclose both the sampling microphone and the voice microphone to isolate sound interference inside the casing; S2. The echo microphone is set at a distance of less than or equal to 5mm from the speaker. The audio signal output by the speaker is collected through the echo microphone as a reference signal for echo cancellation, and the first ambient sound is collected. S3. Set the voice microphone at a distance of 10cm or more from the speaker, and collect the user's voice signal and the second ambient sound through the voice microphone; S4. Based on the reference signal, the speaker echo in the user voice signal is eliminated; S5. Perform a differential operation on the first ambient sound and the second ambient sound to eliminate environmental noise and output a clean user voice signal.

[0008] Preferably, the sealing process involves: creating a sealed cavity on the back of the microphone, wrapping it with sound-absorbing material, or performing a potting process.

[0009] Preferably, before performing the difference operation on the first ambient sound and the second ambient sound, the method further includes: performing delay alignment and amplitude normalization on the first ambient sound and the second ambient sound.

[0010] Secondly, the present invention provides an echo cancellation and environmental noise reduction device based on dual-microphone collaboration, comprising: A sampling microphone is positioned at a distance of less than or equal to 5mm from the speaker and has a first enclosed structure. It is used to collect the audio signal output by the speaker as a reference signal for echo cancellation and to collect the first ambient sound. A voice microphone is positioned at a distance of 10cm or more from the speaker and has a second enclosed structure for collecting user voice signals and second ambient sounds. The signal processing unit is electrically connected to the sampling microphone and the voice microphone, respectively. The signal processing unit is used to perform speaker echo cancellation processing on the user voice signal according to the reference signal, and to perform differential operation on the first ambient sound and the second ambient sound to eliminate ambient noise.

[0011] Preferably, the first sealing structure and the second sealing structure are sealing sleeves, sealing sponges, or sealed cavities.

[0012] Preferably, the signal processing unit includes: The echo cancellation module is used to adaptively filter the user voice signal collected by the voice microphone with reference to the reference signal collected by the echo microphone, so as to cancel the speaker echo. The differential noise reduction module is used to perform differential operations on the first ambient sound collected by the back-sampling microphone and the second ambient sound collected by the voice microphone to eliminate ambient noise.

[0013] Preferably, the signal processing unit is further configured to perform delay alignment and amplitude normalization on the first ambient sound and the second ambient sound before performing differential operation on them.

[0014] Therefore, the present invention employs the above-mentioned method and apparatus for echo cancellation and environmental noise reduction based on dual-microphone collaboration, which has the following beneficial effects: (1) By setting the echo microphone at a distance of less than or equal to 5mm from the speaker, the audio signal output by the speaker can be accurately collected as the reference signal for echo cancellation. This can specifically cancel the speaker echo mixed in the voice microphone, solve the echo interference problem from the source, avoid the defects such as unclear speech and howling caused by echo, and greatly improve the clarity of voice collection and interaction. (2) By synchronously collecting ambient sound through the back sampling microphone and the voice microphone (the back sampling microphone collects the first ambient sound, and the voice microphone collects the second ambient sound), the two ambient sound signals are processed by differential operation, which can effectively cancel out various noises in the environment (such as external human voices, ambient noise, etc.) while retaining the pure user voice signal, thus solving the shortcomings of poor environmental noise reduction effect and voice signal aliasing in the traditional single microphone solution. (3) Both the acquisition microphone and the voice microphone are sealed to effectively isolate the sound interference inside the casing and avoid the noise of the internal circuit and vibration noise affecting the signal acquisition. At the same time, the differentiated layout of "acquisition microphone close to the speaker (≤5mm) and voice microphone far away from the speaker (≥10cm)" further distinguishes the speaker signal and the human voice signal, reduces mutual interference, and significantly improves the overall anti-interference performance. (4) Only two microphones are used in combination with closed processing and differentiated layout. There is no need to add complex acoustic isolation structure or additional noise reduction hardware. The structure is simple and easy to assemble. It can be widely adapted to various electronic devices that require voice acquisition, echo cancellation and environmental noise reduction (such as smart speakers, headphones, conference phones, car microphones, etc.). It has low mass production difficulty, controllable cost and strong practicality.

[0015] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the process of an echo cancellation and environmental noise reduction method based on dual-microphone collaboration according to the present invention; Figure 2 This is a schematic diagram of the layout of the dual microphones relative to the speaker according to an embodiment of the present invention; Figure Labels 1. Speaker; 2. Acoustic microphone; 3. Voice microphone. Detailed Implementation

[0017] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Example like Figure 1 As shown, the present invention provides an echo cancellation and environmental noise reduction method based on dual-microphone collaboration, comprising the following steps: S1. Provide a data acquisition microphone and a voice microphone, and enclose both the data acquisition microphone and the voice microphone to isolate sound interference inside the casing; Specifically, the sealing process can be achieved using any of the following methods: creating a sealed cavity on the back of the microphone, wrapping it with sound-absorbing material, or potting it. This sealing process effectively prevents interference signals such as internal circuit noise and vibration noise from entering the microphone, ensuring the purity of the acquired signal.

[0020] S2. Set the echo microphone at a distance of less than or equal to 5mm from the speaker, and use the echo microphone to collect the audio signal output by the speaker as the reference signal for echo cancellation, and collect the first ambient sound. like Figure 2 As shown, the echo pickup microphone 2 is positioned close to the speaker 1, with a distance of ≤5mm between them. This distance ensures that the signal-to-noise ratio of the speaker 1 output signal acquired by the echo pickup microphone 2 is high, thus serving as an accurate reference for the echo cancellation algorithm. Simultaneously, the echo pickup microphone 2 also synchronously acquires ambient sounds from the surrounding environment (including external human voices, environmental noise, etc.).

[0021] S3. Set the voice microphone at a distance of 10cm or more from the speaker, and collect the user's voice signal and the second ambient sound through the voice microphone; like Figure 2 As shown, the voice microphone 3 is positioned away from the speaker 1, with a distance of ≥10cm between them. This distance effectively prevents direct crosstalk from the audio signal output by the speaker 1 to the user's voice acquisition, resulting in a purer user voice signal acquired by the voice microphone 3. Simultaneously, the voice microphone 3 also acquires ambient sound.

[0022] S4. Based on the reference signal, the speaker echo in the user's voice signal is eliminated; This step uses adaptive filtering technology to achieve echo cancellation. The original signal captured by the voice microphone is... The reference signal acquired by the acquisition microphone is Echo cancellation is performed using a Normalized Least Mean Square (NLMS) adaptive filter, and its core iterative formula is as follows: ; ; ; in, The discrete-time sampling point number; The raw signal captured by the voice microphone (including user voice and speaker echo); The reference signal (i.e., the audio signal output by the speaker) is collected by the sampling microphone. From the present and the past A vector composed of reference signal samples, The filter order; For the first The adaptive filter coefficient vector at time step 1, with length and same; The estimated echo signal is the filter output. The error signal is the user's voice signal after echo cancellation (which still contains ambient noise). This is the step size factor, which controls the convergence speed of the filter. The regularization parameter is a small positive integer used to prevent the denominator from being zero; a typical value is 0.001.

[0023] Using the above iterative formula The user's voice signal is close to pure (affected only by ambient noise), thus eliminating speaker echo.

[0024] S5. Perform differential operation on the first ambient sound and the second ambient sound to eliminate environmental noise and output a clean user voice signal.

[0025] The first ambient sound collected by the microphone is recorded as follows: The second ambient sound collected by the voice microphone is recorded as (here) With S4 It is the same physical signal, namely the total signal collected by the voice microphone, which simultaneously includes the user's voice, speaker echo, and ambient noise. After echo cancellation by S4, the resulting signal has eliminated the echo but still contains ambient noise. To more accurately eliminate environmental noise, differential processing is performed directly using the original two-channel environmental sound information. The following differential noise reduction process is used. First, the two ambient sound channels need to be delayed and normalized in terms of amplitude to ensure the effectiveness of the differential operation.

[0026] Delay alignment: Calculate the cross-correlation function ; in, The first ambient sound signal in the index The sampled value at that location; The second ambient sound signal is in the index The sampled value at that location; The summation index represents the sampling point number, with values ​​ranging from 0 to... .

[0027] Find the delay that maximizes the cross-correlation. And perform delay compensation for the second ambient sound: ; in, For delayed variables (integers); for and The cross-correlation function; The delay value at which the cross-correlation reaches its maximum value; This is the second ambient sound signal after delay compensation.

[0028] Amplitude normalization: Calculate the root mean square value of each signal separately. ; Then calculate the amplitude matching coefficient. And the amplitude of the compensated second ambient sound is scaled: ; in, The length of the signal used to calculate the root mean square (e.g., the number of sampling points in a short time frame). The root mean square value of the first ambient sound; The root mean square value of the second ambient sound after delay compensation; The amplitude matching coefficient is used to adjust the second ambient sound to the same amplitude level as the first ambient sound; This is the second ambient sound signal after amplitude normalization.

[0029] Difference operations: ; in, This is the clean user voice signal output after differential operation (with echo and environmental noise eliminated simultaneously).

[0030] Through the above differential operation, the environmental noise components shared by the two signals cancel each other out, while the user's voice signal unique to the voice microphone is preserved, thus achieving environmental noise cancellation.

[0031] The present invention also provides an echo cancellation and environmental noise reduction device based on dual-microphone collaboration for performing the above-described method. The device includes: a sampling microphone, a voice microphone, and a signal processing unit.

[0032] Echo microphone 2: Positioned at a distance of 5mm or less from the speaker, and equipped with a first enclosed structure. This echo microphone 2 is used to collect the audio signal output from speaker 1 as a reference signal for echo cancellation, and to collect the first ambient sound. For example... Figure 2 As shown, the sampling microphone 2 is placed close to the speaker 1.

[0033] Voice microphone 3: Positioned at a distance of at least 10cm from speaker 1, and equipped with a second enclosed structure. This voice microphone 3 is used to collect user voice signals and ambient sound. (Example:...) Figure 2 As shown, the voice microphone 3 is placed away from the speaker 1.

[0034] The first and second sealing structures can be sealing sleeves, sealing sponges, or sealed cavities. These sealing structures effectively isolate sound interference within the housing, ensuring the purity of the acquired signal.

[0035] Signal processing unit: Electrically connected to the echo microphone 2 and the voice microphone 3 respectively. The signal processing unit is used to cancel the echo from speaker 1 in the user's voice signal according to the reference signal, and to perform differential operation on the first ambient sound and the second ambient sound to eliminate environmental noise.

[0036] In a preferred implementation, the signal processing unit includes: Echo cancellation module: This module adaptively filters the user's voice signal acquired by the voice microphone 3, using the reference signal acquired by the echo microphone 2 as a reference, to cancel out speaker echo. This module can be implemented using an NLMS adaptive filter.

[0037] Differential noise reduction module: This module performs differential calculations on the first ambient sound collected by the retrieval microphone 2 and the second ambient sound collected by the voice microphone 3 to eliminate ambient noise. Before the differential calculation, this module also performs delay alignment and amplitude normalization on both.

[0038] The aforementioned signal processing unit can be a hardware platform such as a digital signal processor (DSP), microcontroller (MCU), field-programmable gate array (FPGA), or application-specific integrated circuit (ASIC), or it can be implemented by a general-purpose processor executing software programs.

[0039] Therefore, this invention employs the aforementioned method and apparatus for echo cancellation and environmental noise reduction based on dual-microphone collaboration. By placing the sampling microphone close to the speaker (≤5mm) and enclosing it, the reference signal for echo cancellation can be accurately acquired. By placing the voice microphone further away from the speaker (≥10cm) and enclosing it, clean user voice and ambient sound can be captured. Adaptive filtering is used to eliminate speaker echo, and differential operations after delay alignment and amplitude normalization are used to eliminate environmental noise, ultimately outputting high-definition user voice. Experiments show that in a typical indoor environment, the echo cancellation of this invention can reach over 35dB, and the environmental noise suppression can reach over 20dB, which is superior to traditional single-microphone solutions.

[0040] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for echo cancellation and ambient noise reduction based on dual microphone cooperation, characterized in that, Includes the following steps: S1. Provide a sampling microphone and a voice microphone, and enclose both the sampling microphone and the voice microphone to isolate sound interference inside the casing; S2. The echo microphone is set at a distance of less than or equal to 5mm from the speaker. The audio signal output by the speaker is collected through the echo microphone as a reference signal for echo cancellation, and the first ambient sound is collected. S3. Set the voice microphone at a distance of 10cm or more from the speaker, and collect the user's voice signal and the second ambient sound through the voice microphone; S4. Based on the reference signal, the speaker echo in the user voice signal is eliminated; S5. Perform a differential operation on the first ambient sound and the second ambient sound to eliminate environmental noise and output a clean user voice signal.

2. The method of claim 1, wherein, The sealing process involves creating a sealed cavity on the back of the microphone, wrapping it with sound-absorbing material, or performing a potting process.

3. The method of claim 1, wherein, Before performing the difference operation on the first ambient sound and the second ambient sound, the method further includes: performing delay alignment and amplitude normalization on the first ambient sound and the second ambient sound.

4. A device for echo cancellation and ambient noise reduction based on dual microphone cooperation, configured to perform the method of any one of claims 1-3, characterized in that, include: The echo microphone is positioned at a distance of less than or equal to 5mm from the speaker and has a first enclosed structure. It is used to collect the audio signal output by the speaker as a reference signal for echo cancellation and to collect the first ambient sound. A voice microphone is positioned at a distance of 10cm or more from the speaker and has a second enclosed structure for collecting user voice signals and second ambient sounds. The signal processing unit is electrically connected to the sampling microphone and the voice microphone respectively. The signal processing unit is used to perform speaker echo cancellation processing on the user voice signal according to the reference signal, and to perform differential operation on the first ambient sound and the second ambient sound to eliminate ambient noise.

5. The apparatus of claim 4, wherein, The first and second sealing structures are sealing sleeves, sealing sponges, or sealed cavities.

6. The apparatus of claim 4, wherein, The signal processing unit includes: The echo cancellation module is used to adaptively filter the user voice signal collected by the voice microphone with reference to the reference signal collected by the echo microphone, so as to cancel the speaker echo. The differential noise reduction module is used to perform differential operations on the first ambient sound collected by the back-sampling microphone and the second ambient sound collected by the voice microphone to eliminate ambient noise.

7. The apparatus of claim 4, wherein, The signal processing unit is further configured to perform delay alignment and amplitude normalization on the first ambient sound and the second ambient sound before performing differential operation on them.

Images