Particle velocity sensor microarray for voice pickup and voice pickup method
By designing a microarray of particle vibration speed sensor, combining MEMS thermal flow and electret sound pressure sensitive components, 360-degree all-round voice pickup is achieved, solving the problems of low signal-to-noise ratio and large aperture of traditional microphone arrays, and improving voice Pick up quality and signal-to-noise ratio.
Patent Information
- Application Number
- CN202011429411.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2020-12-07
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2040-12-07
AI Technical Summary
Traditional microphone arrays are limited by half-wavelength theory when picking far-field voice, with large apertures, many channels, low signal-to-noise ratio, and difficult to achieve 360-degree all-round pickup. The existing particle vibration speed measurement methods are low sensitivity and response band Narrow, difficult to apply in the field of voice interaction technology.
Design a microarray of particle vibration speed sensor, including two groups of symmetrically distributed particle vibration speed sensitive elements and one sound pressure sensitive element, through the MEMS thermal flow measurement mechanism and electret or silicon wheat sound pressure sensitive elements, after combining The terminal acquisition and processing circuit realizes the acquisition and processing of 2-channel particle vibration speed signals and 1-channel sound pressure signals, and performs sound source orientation.
It realizes 360-degree all-round voice pickup with few channels, high signal-to-noise ratio and small size, improves voice pickup quality and signal-to-noise ratio, avoids the problems of the aperture and number of channels of traditional microphone arrays, and provides more efficient Far-field voice pickup solution.
Smart Images

Figure CN112543397B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to a particle velocity sensor microarray for voice pickup and a voice pickup method. Background Art
[0002] In actual complex environments, when a single microphone picks up voice signals, it will inevitably be interfered by the surrounding environment noise, transmission medium noise, room reverberation, and the voices of other speakers, which will seriously affect the voice pickup quality and speech recognition rate when performing far-field pickup. In order to solve the shortcomings of a single microphone, a microphone array is usually used for far-field voice pickup. The microphone array can perform spatial and temporal spectrum processing on sound pressure signals in different spatial directions to achieve functions such as noise suppression, reverberation removal, human voice interference suppression, sound source direction finding, sound source tracking, and array gain. It completes high-quality far-field pickup at the front end of voice interaction to improve the speech recognition rate in real environments. Traditional microphone arrays are subject to the limitations of half-wavelength theory. The more microphones there are, the larger the aperture. In addition, there are shortcomings such as spatial aliasing and high computational complexity, which greatly limits the design freedom and application scenarios of microphone arrays.
[0003] The speech sound field has both scalar field (sound pressure) and vector field (particle velocity) characteristics, and both sound pressure and particle velocity carry rich speech information. Existing microphone arrays are all based on sound pressure microphones, such as the common MEMS microphones and microphone heads on the market; At present, there are two main measurement methods for speech particle velocity measurement. One is an indirect measurement method, which uses two sound pressure sensors with a certain distance to form a sound pressure gradient to calculate the particle velocity (patent number: 201310726022), but it is limited by the amplitude, phase-frequency consistency and physical distance between the sound pressure microphones, and has the disadvantages of low sensitivity, narrow response frequency band, and large measurement error; the other is a direct measurement method, which is based on the MEMS thermal flow measurement mechanism (patent number: 201310752209). The air sound particle velocity sensitive element has directional, "8"-shaped directional characteristics that do not change with frequency, and directional gain. It has been widely used in advanced individual equipment, sniper positioning, noise source identification and passive acoustic radar, but has not yet been applied to the field of speech interaction technology. Summary of the invention
[0004] The object of the present invention is to provide a particle velocity sensor microarray for voice pickup and a voice pickup method, which has a small number of channels, a high signal-to-noise ratio, and a small size, and can effectively achieve 360-degree omnidirectional voice pickup.
[0005] Based on the same inventive concept, the present invention has two independent technical solutions:
[0006] 1. A particle velocity sensor microarray for voice pickup, comprising a particle velocity sensitive element and a sound pressure sensitive element, comprising two groups of particle velocity sensitive elements, each group of particle velocity sensitive elements is symmetrically distributed, the two groups of particle velocity sensitive elements have the same symmetry center, the two groups of particle velocity sensitive elements are vertically distributed, and the sound pressure sensitive element is located at the symmetry center; the signal output ends of the two groups of particle velocity sensitive elements and the signal output end of the sound pressure sensitive element are connected to the signal input end of a back-end acquisition and processing circuit.
[0007] Furthermore, the two groups of particle velocity sensitive elements and sound pressure sensitive elements are arranged on a microarray skeleton, the microarray skeleton is a regular tetrahedral prism, and the two groups of particle velocity sensitive elements are symmetrically distributed on the sides of the regular tetrahedral prism; the sound pressure sensitive elements are distributed on the upper end surface of the regular tetrahedral prism, and the center of the sound pressure sensitive element is the same as the symmetry center of the regular tetrahedral prism.
[0008] Furthermore, the two groups of particle velocity sensitive elements and sound pressure sensitive elements are arranged on a microarray skeleton, the microarray skeleton is a planar skeleton, the two groups of particle velocity sensitive elements are distributed in a cross shape, and the sound pressure sensitive element is located at the center of the cross.
[0009] Furthermore, a protective cover is arranged outside the microarray framework.
[0010] Furthermore, one sound pressure sensitive element is provided; each group of particle velocity sensitive elements is provided with 2n particle velocity sensitive elements, n≥1, and the size of n is determined according to the signal-to-noise ratio and size of the particle velocity sensor microarray.
[0011] Furthermore, the particle velocity sensing elements on the same side of each group of particle velocity sensing elements are distributed in parallel and at equal intervals.
[0012] Furthermore, the particle velocity sensing element is a particle velocity sensing element based on a MEMS thermal flow measurement mechanism.
[0013] Furthermore, the sound pressure sensitive element is an electret or silicon microphone sound pressure sensitive element.
[0014] 2. A voice pickup method using the above-mentioned particle velocity sensor microarray for voice pickup, wherein the signal output ends of the two groups of particle velocity sensitive elements and the signal of the sound pressure sensitive elements respectively output 2-channel particle velocity signals and 1-channel sound pressure signals, and the back-end acquisition and processing circuit performs sound source orientation based on the input 2-channel particle velocity signals and 1-channel sound pressure signals.
[0015] Furthermore, the back-end acquisition and processing circuit first performs symmetrical simulation summation on the particle velocity analog signals output by the two groups of particle velocity sensitive elements, and then performs sound source orientation based on the trigonometric function relationship between the 2-channel particle velocity signal and the 1-channel sound pressure signal.
[0016] The present invention has the beneficial effects:
[0017] The present invention includes particle velocity sensing elements and sound pressure sensing elements, including two groups of particle velocity sensing elements, each group of particle velocity sensing elements is symmetrically distributed, the two groups of particle velocity sensing elements have the same symmetry center, the two groups of particle velocity sensing elements are perpendicularly distributed to each other, and the sound pressure sensing elements are located at the symmetry center. The present invention has a small number of channels, with one sound pressure channel and two particle velocity channels; it has a small size, a compact microarray structure, and densely arranged particle velocity sensing elements and sound pressure sensing elements; the signal-to-noise ratio can be effectively improved by adding particle velocity sensing elements. The present invention has a small number of channels, a high signal-to-noise ratio, and a small size, and effectively realizes 360-degree omnidirectional voice pickup.
[0018] In the present invention, two groups of particle velocity sensing elements and sound pressure sensing elements are arranged on a microarray skeleton, the microarray skeleton is a regular tetrahedral prism, and the two groups of particle velocity sensing elements are symmetrically distributed on the sides of the regular tetrahedral prism; the sound pressure sensing element is distributed on the upper end surface of the regular tetrahedral prism, and the center of the sound pressure sensing element is the same as the symmetry center of the regular tetrahedral prism; two groups of particle velocity sensing elements and sound pressure sensing elements are arranged on the microarray skeleton, the microarray skeleton is a plane skeleton, the two groups of particle velocity sensing elements are distributed in a cross shape, and the sound pressure sensing element is located at the center of the cross. The present invention adopts the above-mentioned microarray arrangement method, which has a compact structure and a small size, and at the same time ensures a 360-degree omnidirectional voice pickup effect.
[0019] The present invention is provided with one sound pressure sensitive element; each group of particle velocity sensitive elements is provided with 2n particle velocity sensitive elements, and the size of n is determined according to the signal-to-noise ratio and size of the particle velocity sensor microarray. For each doubling of the particle velocity sensitive elements, the signal-to-noise ratio of the microarray is increased by 3dB for spatially uncorrelated Gaussian white noise. Within the allowable size range, increasing the particle velocity sensitive elements can effectively improve the signal-to-noise ratio of the microarray of the present invention.
[0020] The particle velocity sensing element of the present invention is a particle velocity sensing element based on the MEMS thermal flow measurement mechanism; the sound pressure sensing element is an electret or silicon microphone sound pressure sensing element; the signal output ends of the two groups of particle velocity sensing elements and the signal output ends of the sound pressure sensing elements are connected to the signal input end of the back-end acquisition and processing circuit, and the back-end acquisition and processing circuit first symmetrically simulates and sums the particle velocity analog signals output by the two groups of particle velocity sensing elements, and then performs sound source orientation based on the trigonometric function relationship between the 2-channel particle velocity signals and the 1-channel sound pressure signals. The present invention collects voice signals through particle velocity sensing elements and sound pressure sensing elements, and then performs sound source orientation according to the sound pressure signals and particle velocity signals through the back-end acquisition and processing circuit. Since the sound source orientation algorithm and beamforming algorithm are independent of the array receiving frequency and aperture, this is essentially different from the traditional sound pressure microphone array, and the particle velocity sensor microarray size can be millimeter-level (volume less than 1cm 3 ), and the sound source directional accuracy is high; within the range allowed by the size, by increasing the number of particle velocity sensitive elements, the voice pickup signal-to-noise ratio can be improved without causing changes in the number of microarray output channels, thereby solving the problems of large aperture size, large number of channels, and low signal-to-noise ratio of traditional microphone arrays, and providing a better solution for far-field voice pickup. BRIEF DESCRIPTION OF THE DRAWINGS
[0021] Figure 1 is a schematic diagram of the three-dimensional structure of the first embodiment of the present invention;
[0022] Figure 2 is a circuit connection diagram of Embodiment 1 of the present invention;
[0023] Figure 3 It is a circuit connection diagram of the second embodiment of the present invention. DETAILED DESCRIPTION
[0024] The present invention is described in detail below with reference to the embodiments shown in the accompanying drawings, but it should be noted that these embodiments are not limitations of the present invention, and any equivalent transformations or substitutions in functions, methods, or structures made by ordinary technicians in the field based on these embodiments are all within the scope of protection of the present invention.
[0025] Embodiment 1:
[0026] Particle velocity sensor microarray
[0027] like Figure 1 , Figure 2As shown, the particle velocity sensor microarray includes particle velocity sensitive elements and sound pressure sensitive elements 2, including two groups of particle velocity sensitive elements, namely a first group of particle velocity sensitive elements 11 and a second group of particle velocity sensitive elements 12, each group of particle velocity sensitive elements is symmetrically distributed, the symmetry centers of the two groups of particle velocity sensitive elements are the same, the two groups of particle velocity sensitive elements 11 and 12 are perpendicularly distributed to each other, and the sound pressure sensitive element 2 is located at the symmetry center. In this embodiment, the two groups of particle velocity sensitive elements 11 and 12 and the sound pressure sensitive element 2 are arranged on a microarray skeleton 31, and the microarray skeleton 31 is a regular tetrahedral prism, and the two groups of particle velocity sensitive elements 11 and 12 are symmetrically distributed on the side of the regular tetrahedral prism; the sound pressure sensitive element 2 is distributed on the upper end surface of the regular tetrahedral prism, and the center of the sound pressure sensitive element 2 is the same as the symmetry center of the regular tetrahedral prism. There is one sound pressure sensitive element 2; each group of particle velocity sensitive elements is provided with 2n particle velocity sensitive elements, n≥1, and the size of n is determined according to the signal-to-noise ratio and size of the particle velocity sensor microarray. The particle velocity sensitive elements on the same side of each group of particle velocity sensitive elements are equally spaced and parallelly distributed, with a spacing of d. The particle velocity sensitive elements 11 and 12 are particle velocity sensitive elements based on the MEMS thermal flow measurement mechanism. The sound pressure sensitive element 2 is an electret, a silicon wheat sound pressure sensitive element or other types of sound pressure sensitive elements. A protective cover 4 is provided on the outside of the microarray skeleton.
[0028] like Figure 2 As shown, the signal output terminals Sig1 and Sig2 of the two groups of particle velocity sensing elements and the signal output terminal Sig3 of the sound pressure sensing element are connected to the signal input terminal of the back-end acquisition and processing circuit.
[0029] Embodiment 2:
[0030] Particle velocity sensor microarray
[0031] like Figure 3 As shown, in the second embodiment, two groups of particle velocity sensors 11, 12 and a sound pressure sensor 2 are arranged on a microarray skeleton 32, the microarray skeleton 32 is a planar skeleton, the two groups of particle velocity sensors 11, 12 are distributed in a cross shape, and the sound pressure sensor 2 is located at the center of the cross. The rest of the structure and working principle of the second embodiment are the same as those of the first embodiment.
[0032] Embodiment three:
[0033] A method for voice pickup using the particle velocity sensor microarray
[0034] The back-end acquisition and processing circuit first performs symmetrical simulation summation on the particle velocity analog signals output by the two groups of particle velocity sensitive elements 11 and 12, and then performs sound source orientation based on the trigonometric function relationship between the 2-channel particle velocity signal and the 1-channel sound pressure signal.
[0035] When working, the microarray outputs a total of three-channel signals, namely, the first channel particle velocity signal output by the first group of particle velocity sensitive elements 11 from the signal output terminal Sig1; the second channel particle velocity signal output by the second group of particle velocity sensitive elements 12 from the signal output terminal Sig2; the sound pressure signal output by the sound pressure sensitive element 2 from the signal output terminal Sig3. The back-end acquisition and processing circuit symmetrically simulates and sums the first channel particle velocity signal and the second channel particle velocity signal respectively. After the symmetrical simulation and summation, the two channel particle velocity signals are horizontally co-pointed and orthogonal, so the present invention effectively reduces the number of channels. The back-end acquisition and processing circuit is based on the trigonometric function relationship between the 2-channel particle velocity signal and the 1-channel sound pressure signal. When performing sound source orientation, the sound source azimuth DOA (Direction of arrival) estimation method such as the complex sound intensity method and the histogram method can be used.
[0036] The trigonometric function relationship between the above-mentioned 1-channel sound pressure signal and 2-channel particle velocity signal is determined by the omnidirectional directivity of the sound pressure sensitive element and the "8"-shaped spatial directivity of the particle velocity sensitive element, and has nothing to do with the receiving frequency f and array aperture d of the sound pressure sensitive element and the particle velocity sensitive element. The 1-channel sound pressure signal and the 2-channel particle velocity signal that are orthogonal to the horizontal common point can be combined with the electronic rotation method to form a "single-sided" directional beam pointing to the sound source. Therefore, within the allowable range of the microarray size, the array size can be achieved at the millimeter level. The signal-to-noise ratio of the entire microarray can be improved by increasing the number of the particle velocity sensitive elements 1. Specifically, when the number of the first group of particle velocity sensitive elements 11 and the second group of particle velocity sensitive elements 12 is doubled, the signal-to-noise ratio of the microarray can be improved by 3dB for spatially uncorrelated Gaussian white noise. The implementation process is as follows:
[0037] For spatially uncorrelated Gaussian white noise, the array gain of the microarray is (reference Huang Geng, Zhang Bei; Simulation study on array gain performance of common arrays based on Matlab [J]):
[0038] G(dB)=10log(N)
[0039] Wherein, N is a multiple of the number of particle velocity sensitive elements 1 in the microarray.
[0040] When the number of the particle velocity sensitive elements 1 is doubled, the array gain of the microarray is increased by:
[0041] △G(dB)=10log(2N)-10log(N)=10log(2)=3dB
[0042] Although the number of the particle velocity sensitive elements 1 is increased, the number of output channels of the microarray does not change, which is an advantage over the traditional microphone array. Since the microarray sound source orientation is independent of the receiving frequency f of the sound pressure sensitive element 2 and the particle velocity sensitive element 1 and the array aperture d, the first group of particle velocity sensitive elements 11 and the second group of particle velocity sensitive elements 12 can be arranged more densely. When n=1, the volume of the microarray is only 0.5cm 3 When the size of the microarray is fixed, the number of the particle velocity sensitive elements 1 can be increased as much as possible to obtain a higher signal-to-noise ratio.
[0043] The series of detailed descriptions listed above are only specific descriptions of feasible implementation methods of the present invention. They are not intended to limit the scope of protection of the present invention. Any equivalent implementation methods or changes that do not deviate from the technical spirit of the present invention should be included in the scope of protection of the present invention.
[0044] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the present invention can be implemented in other specific forms without departing from the spirit or essential features of the present invention. Therefore, the embodiments should be considered exemplary and non-restrictive in all respects, and the scope of the present invention is defined by the appended claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims be included in the present invention.
Claims
1. A particle velocity sensor microarray for voice pickup, comprising a particle velocity sensing element and a sound pressure sensing element (2), characterized in that: It comprises two groups of particle velocity sensing elements (11, 12), each group of particle velocity sensing elements is symmetrically distributed, the symmetry centers of the two groups of particle velocity sensing elements are the same, the two groups of particle velocity sensing elements are vertically distributed, and the sound pressure sensing element (2) is located at the symmetry center; the signal output ends of the two groups of particle velocity sensing elements (11, 12) and the signal output end of the sound pressure sensing element (2) are connected to the signal input end of a back-end acquisition and processing circuit; The two groups of particle velocity sensing elements (11, 12) and sound pressure sensing elements (2) are arranged on a microarray skeleton (31); The microarray skeleton (31) is a regular tetrahedral prism, the two groups of particle velocity sensing elements (11, 12) are symmetrically distributed on the side surfaces of the regular tetrahedral prism; the sound pressure sensing element (2) is distributed on the upper end surface of the regular tetrahedral prism, and the center of the sound pressure sensing element (2) is the same as the symmetry center of the regular tetrahedral prism; or The microarray skeleton (32) is a planar skeleton, the two groups of particle velocity sensitive elements are distributed in a cross shape, and the sound pressure sensitive element is located at the center of the cross; There is one sound pressure sensitive element; each group of particle velocity sensitive elements is provided with 2n particle velocity sensitive elements, n≥1, and the size of n is determined according to the signal-to-noise ratio and size of the particle velocity sensor microarray.
2. The particle velocity sensor microarray for voice pickup according to claim 1, characterized in that: A protective cover (4) is arranged outside the microarray framework.
3. The particle velocity sensor microarray for voice pickup according to claim 1, characterized in that: The particle velocity sensing elements on the same side of each group of particle velocity sensing elements are distributed in parallel with equal spacing.
4. The particle velocity sensor microarray for voice pickup according to claim 1, characterized in that: The particle vibration velocity sensing elements (11, 12) are particle vibration velocity sensing elements based on a MEMS thermal flow measurement mechanism.
5. The particle velocity sensor microarray for voice pickup according to claim 1, characterized in that: The sound pressure sensitive element (2) is an electret or silicon microphone sound pressure sensitive element.
6. A method for voice pickup using the particle velocity sensor microarray for voice pickup according to any one of claims 1 to 5, characterized in that: The signal output ends of the two groups of particle velocity sensing elements (11, 12) and the signal of the sound pressure sensing element (2) respectively output two-channel particle velocity signals and one-channel sound pressure signal, and the back-end acquisition processing circuit performs sound source orientation according to the input two-channel particle velocity signals and one-channel sound pressure signal.
7. The voice pickup method according to claim 6, characterized in that: The back-end acquisition and processing circuit first symmetrically simulates and sums the particle velocity analog signals output by the two groups of particle velocity sensitive elements, and then performs sound source orientation based on the trigonometric function relationship between the 2-channel particle velocity signal and the 1-channel sound pressure signal.
