CN105025416A - Portable double-microphone sound source identification and positioning device - Google Patents

Abstract

The invention relates to a sound source identification and positioning device and especially relates to a portable double-microphone sound source identification and positioning device. The technical problem to be solved is to provide the portable double-microphone sound source identification and positioning device which is simple in structure, small in size, convenient to carry and more visual in display result. The portable double-microphone sound source identification and positioning device comprises a microphone A, a microphone B, a pre-amplifier A, a pre-amplifier B, a support, a signal acquisition module, a sound source identification and positioning module, a storage module and a displayer. The portable double-microphone sound source identification and positioning device solves the defects that an existing sound source identification and positioning device is large in size, and inconvenient to transport and carry; the device is simple in structure, small in size, convenient to carry, capable of identifying sound sources from different directions, small in computation complexity, high in real-time performance, and suitable for being applied to an embedded system; and the device is equipped with a liquid crystal display screen, so that the display results are more visual.

Description

A kind of portable two microphones Sound sources identification and localization device

Technical field

The present invention relates to a kind of Sound sources identification and localization device, particularly relate to a kind of portable two microphones Sound sources identification and localization device.

Background technology

In intelligent monitoring and mobile robot's research field, the separation of sound-source signal is the focus studied with location always, and the identification how making robot only utilize two microphones to complete sound source as people is the difficult point studied with location; Under numerous Conversation Voice and noisy environment, Sound sources identification and localization technology can focus on the echo signal interested to user, suppresses interference noise, contributes to the separation of PMD EDM; In addition, deft design, high accuracy and the sound source direction device that is convenient for carrying in medical treatment, search and rescue, the numerous areas such as military and national defence has important using value.

The microphone array array element number that existing sound source direction device adopts is many, bulky, only has large-scale vehicle-mounted sound source sniffer, does not have small-sized portable sound source sniffer.Patent CN101910807A uses the sound source identifying and measuring apparatus possessing baffle plate to obtain all directions sound source information, and by the azimuth of passing generation in time, the elevation angle, acoustic pressure, frequency frequency characteristic etc. connects, determine sound source more accurately, but microphone array is bulky, data operation quantity is large, be not suitable for real-time process, patent CN104049235A discloses a kind of sound source direction device, microphone array comprises four array elements, complex structure, repeatedly need adjust and determine target bearing, patent CN101957442B adopts the microphone array of quaternary positive tetrahedron structure, four microphones lay respectively at four summits of this positive tetrahedron structure, institute takes up space larger, be inconvenient to transport and carry, patent CN103219012A utilizes dual microphone noise reduction, strengthen voice, but do not provide the azimuth information of sound source.

Summary of the invention

(1) technical problem that will solve

The present invention is large in order to overcome existing Sound sources identification and localization device volume, inconvenience transport and the shortcoming of carrying, it is simple that the technical problem to be solved in the present invention is to provide a kind of device structure, and volume is little, be easy to carry, the more intuitive portable two microphones Sound sources identification and localization device of display result.

(2) technical scheme

In order to solve the problems of the technologies described above, the invention provides so a kind of portable two microphones Sound sources identification and localization device, comprise microphone A, microphone B, preamplifier A, preamplifier B, support, signal acquisition module, Sound sources identification and localization module, memory module, display; Described microphone A is connected on preamplifier A, described microphone B is connected on preamplifier B, described microphone A, microphone B are connected on support, arrange in H type, described preamplifier A, preamplifier B are connected to signal acquisition module, described signal acquisition module, memory module, display are connected in Sound sources identification and localization module, and microphone A and microphone B periphery are all enclosed with electromic thermostat.

Preferably, described microphone A is connected with preamplifier A by screw thread, and microphone B is connected with preamplifier B by screw thread.

Preferably, the directive property of described microphone A and described microphone B is omni-directional.

Preferably, described support is " Н " type, and wherein two is circular groove, and described microphone A and described microphone B embeds support both sides respectively, and keeping parallelism is arranged.

Preferably, long 5 centimetres of horizontal edge in the middle of described " Н " type support.

Preferably, described preamplifier A and described preamplifier B is built-in.

Preferably, described display is liquid crystal display.

Preferably, described memory module stores the institute's sound source obtained by Sound sources identification and localization module, and described display shows amplitude and the phase place of institute's sound source by polar diagram.

Operation principle: first utilize microphone A and microphone B to gather sound-source signal respectively, and acoustical signal is converted into the signal of telecommunication; Then by preamplifier A and preamplifier B, after respectively the signal of telecommunication being amplified, signal acquisition module is transferred to; Two-way acoustical signal is carried out filtering, analog/digital conversion by signal acquisition module, and the two ways of digital signals after conversion transfers to Sound sources identification and localization module; In Sound sources identification and localization module, utilize normalization, first-order difference, time-frequency divides, isolated component scheduling algorithm carries out identification and the location of sound source; Position amplitude and the phase information of sound source is shown in last liquid crystal display.

In Sound sources identification and localization module, utilize normalization, first-order difference, time-frequency divides, isolated component scheduling algorithm carries out identification and the location of sound source.Method wherein based on independent component analysis processes signal, and concrete steps are as follows:

Catabolic phase: utilize Computational auditory scene analysis to be separated two-way collection signal, obtains more independently two separation components;

Organize the stage: two components are converted into time-frequency domain by time domain, are decomposed into basic time frequency unit, calculate the time-frequency mask of each time frequency unit, according to time-frequency mask, the time frequency unit with same characteristic features is organized together formation sound clip;

Merging phase: merge the sound clip with similar features, reconstruct time domain waveform by time-frequency domain fragment, judging that whether this synthesis sound source is from same sound source, if so, is then final separation component; If not, then return and organize the stage, continue to be separated till satisfying condition;

Positioning stage: utilize time delay and phase difference to position sound source, first utilizes filter to carry out preliminary treatment to noise in voice signal, then estimates the time delay of multi-path voice signal, finally builds the position that geometrical relationship determines sound source.

In catabolic phase, Computational auditory scene analysis method adopts Informax Independent Component Analysis Algorithm or combines with time-frequency mask based on the joint approximate diagonalization algorithm of tensor;

Organize in the stage and two components are converted into time-frequency domain by time domain, adopt the one in Short Time Fourier Transform, wavelet transformation and the filtering of Gammatone bank of filters to convert component, obtain basic time frequency unit representation.

Calculate masking factor to all time frequency unit in merging phase, build time-frequency masking matrix, retain the time frequency unit of target sound source, the time frequency unit of the interference signal that significantly decays, building formula is:

$M_1(\mathrm{\ω},t)=\left\{\begin{array}{cc}1,& if\left|Y_1(\mathrm{\ω},t)\right|>\mathrm{\τ}\left|Y_2(\mathrm{\ω},t)\right|\\ 0,& \mathrm{oth}erwise\∀\mathrm{\ω},t\end{array}\right.$

$M_2(\mathrm{\ω},t)=\left\{\begin{array}{cc}1,& if\left|Y_2(\mathrm{\ω},t)\right|>\mathrm{\τ}\left|Y_1(\mathrm{\ω},t)\right|\\ 0,& otherwise\∀\mathrm{\ω},t\end{array}\right.$

In formula, Y ₁(ω, t), Y ₂(ω, t) is the time-frequency representation of two components, and τ is degree of rarefication controling parameters.

Organize and the time frequency unit with same characteristic features organized together according to time-frequency mask the formula forming sound clip in the stage and be:

$\left.\begin{array}{c}{x}_{11}(\mathrm{\ω},t)\\ x_{12}(\mathrm{\ω},t)\end{array}\right\}={\mathrm{BM}}_1(\mathrm{\ω},t)\left\{\begin{array}{c}{X}_1(\mathrm{\ω},t)\\ X_2(\mathrm{\ω},t)\end{array}\right.$

$\left.\begin{array}{c}{x}_{21}(\mathrm{\ω},t)\\ x_{22}(\mathrm{\ω},t)\end{array}\right\}={\mathrm{BM}}_2(\mathrm{\ω},t)\left\{\begin{array}{c}{X}_1(\mathrm{\ω},t)\\ X_2(\mathrm{\ω},t)\end{array}\right.$

In formula, X ₁(ω, t), X ₂(ω, t) is the time-frequency representation of microphone collection signal.

The method of rebuilding time domain waveform in merging phase is corresponding with the time-frequency conversion method that merging phase adopts, and adopts inverse Fourier transform, wavelet reconstruction and the one in Gammatone bank of filters in short-term.

Judge described in merging phase that whether synthesis sound source is from same sound source, basis for estimation is the conditional number of covariance matrix, definition covariance matrix:

$R_{xx}=<{\mathrm{xx}}^T>=\frac{1}{N_s}{\mathrm{xx}}^T$

In formula, x is the time-domain signal of synthesis sound source, N _sfor the sampling number of x, Matrix condition number computing formula is:

cond(R _xx)＝maxeig(R _xx)/mineig(R _xx)

Wherein, all characteristic values of eig () representing matrix, when conditional number is greater than certain threshold value, judge that output signal is from same sound source.

And, we have installed electromic thermostat additional in the periphery of microphone, two microphones are made to keep identical and stationary temperature, general control is at 20 degree, decrease the temperature drift phenomenon of microphone, find, after having installed electromic thermostat additional through overtesting, around when ambient temperature too high (being greater than 30 degree) or too low (being less than 10 degree), be greatly improved than the accuracy of the accuracy rate and location that do not install electromic thermostat identification of sound source additional.

(3) beneficial effect

One provided by the invention portable two microphones Sound sources identification and localization device, overcomes existing Sound sources identification and localization device volume large, inconvenience transport and the shortcoming of carrying, and only use two microphones, device structure is simple, and volume is little, is easy to carry; The recognition methods that independent component analysis combines with time-frequency mask is not done sound source number and route of transmission and is supposed, accurately can identify that amount of calculation is little from different azimuth sound source, ensure that the requirement of real-time of sonic location system, is applicable to the application of embedded system; Be configured with LCDs, and utilize polar form to show each individual sources, display result is more directly perceived.

Accompanying drawing explanation

Fig. 1 is structural representation of the present invention.

Fig. 2 is structural representation of the present invention.

Fig. 3 is signals collecting schematic diagram of the present invention.

Fig. 4 is sound source position of the present invention display schematic diagram.

Fig. 5 is the identification of sound source and the separation process figure that the present invention is based on isolated component.

Being labeled as in accompanying drawing: 1-microphone A, 2-microphone B, 3-preamplifier A, 4-preamplifier B, 5-signal acquisition module, 6-Sound sources identification and localization module, 7-display, 8-memory module, 9-support, 10-electromic thermostat.

Embodiment

Below in conjunction with drawings and Examples, the present invention is further illustrated.

Embodiment 1

A kind of portable two microphones Sound sources identification and localization device, as shown in Figure 1-2, microphone A1, microphone B2, preamplifier A3, preamplifier B4, support 9, signal acquisition module 5, Sound sources identification and localization module 6, memory module 8, display 7 is comprised; Described microphone A1 is connected on preamplifier A3, described microphone B2 is connected on preamplifier B4, described microphone A1, microphone B2 are connected on support 9, arrange in H type, described preamplifier A3, preamplifier B4 are connected to signal acquisition module 5, described signal acquisition module 5, memory module 8, display 7 are connected in Sound sources identification and localization module 6, and microphone A1 and microphone B2 periphery are all enclosed with electromic thermostat 10.

Operation principle:

We have installed electromic thermostat 10 additional in the periphery of microphone, two microphones are made to keep identical and stationary temperature, general control is at 20 degree, decrease the temperature drift phenomenon of microphone, find through overtesting, after having installed electromic thermostat 10 additional, around when ambient temperature too high (being greater than 30 degree) or too low (being less than 10 degree), be greatly improved than the accuracy of the accuracy rate and location that do not install electromic thermostat 10 identification of sound source additional.

Embodiment 2

A kind of portable two microphones Sound sources identification and localization device, as Figure 1-5, microphone A1, microphone B2, preamplifier A3, preamplifier B4, support 9, signal acquisition module 5, Sound sources identification and localization module 6, memory module 8, display 7 is comprised; Described microphone A1 is connected on preamplifier A3, described microphone B2 is connected on preamplifier B4, described microphone A1, microphone B2 are connected on support 9, arrange in H type, described preamplifier A3, preamplifier B4 are connected to signal acquisition module 5, described signal acquisition module 5, memory module 8, display 7 are connected in Sound sources identification and localization module 6, and microphone A1 and microphone B2 periphery are all enclosed with electromic thermostat 10.

Described microphone A1 is connected with preamplifier A3 by screw thread, and microphone B2 is connected with preamplifier B4 by screw thread.

The directive property of described microphone A1 and described microphone B2 is omni-directional.

Described support 9 is " Н " type, and wherein two is circular groove, and described microphone A1 and described microphone B2 embeds support 9 both sides respectively, and keeping parallelism is arranged.

Long 5 centimetres of the middle horizontal edge of described " Н " type support 9.

Described preamplifier A3 and described preamplifier B4 is built-in.

Described display 7 is liquid crystal display 7.

Described memory module 8 stores the institute's sound source obtained by Sound sources identification and localization module 6, and described display 7 shows amplitude and the phase place of institute's sound source by polar diagram.

Operation principle: first utilize microphone A1 and microphone B2 to gather sound-source signal respectively, and acoustical signal is converted into the signal of telecommunication; Then by preamplifier A3 and preamplifier B4, signal acquisition module 5 after respectively the signal of telecommunication being amplified, is transferred to; Two-way acoustical signal is carried out filtering, analog/digital conversion by signal acquisition module 5, and the two ways of digital signals after conversion transfers to Sound sources identification and localization module 6; In Sound sources identification and localization module 6, utilize normalization, first-order difference, time-frequency divides, isolated component scheduling algorithm carries out identification and the location of sound source; Position amplitude and the phase information of sound source is shown in last liquid crystal display 7.

In Sound sources identification and localization module 6, utilize normalization, first-order difference, time-frequency divides, isolated component scheduling algorithm carries out identification and the location of sound source.Method wherein based on independent component analysis processes signal, and concrete steps are as follows:

Catabolic phase: utilize Computational auditory scene analysis to be separated two-way collection signal, obtains more independently two separation components;

Organize the stage: two components are converted into time-frequency domain by time domain, are decomposed into basic time frequency unit, calculate the time-frequency mask of each time frequency unit, according to time-frequency mask, the time frequency unit with same characteristic features is organized together formation sound clip;

Merging phase: merge the sound clip with similar features, reconstruct time domain waveform by time-frequency domain fragment, judging that whether this synthesis sound source is from same sound source, if so, is then final separation component; If not, then return and organize the stage, continue to be separated till satisfying condition;

Positioning stage: utilize time delay and phase difference to position sound source, first utilizes filter to carry out preliminary treatment to noise in voice signal, then estimates the time delay of multi-path voice signal, finally builds the position that geometrical relationship determines sound source.

In catabolic phase, Computational auditory scene analysis method adopts Informax Independent Component Analysis Algorithm or combines with time-frequency mask based on the joint approximate diagonalization algorithm of tensor;

Organize in the stage and two components are converted into time-frequency domain by time domain, adopt the one in Short Time Fourier Transform, wavelet transformation and the filtering of Gammatone bank of filters to convert component, obtain basic time frequency unit representation.

Calculate masking factor to all time frequency unit in merging phase, build time-frequency masking matrix, retain the time frequency unit of target sound source, the time frequency unit of the interference signal that significantly decays, building formula is:

$M_1(\mathrm{\&omega;},t)=\left\{\begin{array}{cc}1,& if\left|Y_1(\mathrm{\&omega;},t)\right|>\mathrm{\&tau;}\left|Y_2(\mathrm{\&omega;},t)\right|\\ 0,& otherwise\&ForAll;\mathrm{\&omega;},t\end{array}\right.$

$M_2(\mathrm{\&omega;},t)=\left\{\begin{array}{cc}1,& if\left|Y_2(\mathrm{\&omega;},t)\right|>\mathrm{\&tau;}\left|Y_1(\mathrm{\&omega;},t)\right|\\ 0,& otherwise\&ForAll;\mathrm{\&omega;},t\end{array}\right.$

In formula, Y ₁(ω, t), Y ₂(ω, t) is the time-frequency representation of two components, and τ is degree of rarefication controling parameters.

Organize and the time frequency unit with same characteristic features organized together according to time-frequency mask the formula forming sound clip in the stage and be:

$\left.\begin{array}{c}{x}_{11}(\mathrm{\&omega;},t)\\ x_{12}(\mathrm{\&omega;},t)\end{array}\right\}={\mathrm{BM}}_1(\mathrm{\&omega;},t)\left\{\begin{array}{c}{X}_1(\mathrm{\&omega;},t)\\ X_2(\mathrm{\&omega;},t)\end{array}\right.$

$\left.\begin{array}{c}{x}_{21}(\mathrm{\&omega;},t)\\ x_{22}(\mathrm{\&omega;},t)\end{array}\right\}={\mathrm{BM}}_2(\mathrm{\&omega;},t)\left\{\begin{array}{c}{X}_1(\mathrm{\&omega;},t)\\ X_2(\mathrm{\&omega;},t)\end{array}\right.$

In formula, X ₁(ω, t), X ₂(ω, t) is the time-frequency representation of microphone collection signal.

The method of rebuilding time domain waveform in merging phase is corresponding with the time-frequency conversion method that merging phase adopts, and adopts inverse Fourier transform, wavelet reconstruction and the one in Gammatone bank of filters in short-term.

Judge described in merging phase that whether synthesis sound source is from same sound source, basis for estimation is the conditional number of covariance matrix, definition covariance matrix:

$R_{xx}=<{\mathrm{xx}}^T>=\frac{1}{N_s}{\mathrm{xx}}^T$

In formula, x is the time-domain signal of synthesis sound source, N _sfor the sampling number of x, Matrix condition number computing formula is:

cond(R _xx)＝maxeig(R _xx)/mineig(R _xx)

Wherein, all characteristic values of eig () representing matrix, when conditional number is greater than certain threshold value, judge that output signal is from same sound source.

And, we have installed electromic thermostat 10 additional in the periphery of microphone, two microphones are made to keep identical and stationary temperature, general control is at 20 degree, decrease the temperature drift phenomenon of microphone, find, after having installed electromic thermostat 10 additional through overtesting, around when ambient temperature too high (being greater than 30 degree) or too low (being less than 10 degree), be greatly improved than the accuracy of the accuracy rate and location that do not install electromic thermostat 10 identification of sound source additional.

The above embodiment only have expressed the preferred embodiment of the present invention, and it describes comparatively concrete and detailed, but therefore can not be interpreted as the restriction to the scope of the claims of the present invention.It should be pointed out that for the person of ordinary skill of the art, without departing from the inventive concept of the premise, can also make some distortion, improvement and substitute, these all belong to protection scope of the present invention.Therefore, the protection range of patent of the present invention should be as the criterion with claims.

Claims (8)

1. a portable two microphones Sound sources identification and localization device, it is characterized in that, comprise microphone A (1), microphone B (2), preamplifier A (3), preamplifier B (4), support (9), signal acquisition module (5), Sound sources identification and localization module (6), memory module (8), display (7), described microphone A (1) is connected on preamplifier A (3), described microphone B (2) is connected on preamplifier B (4), described microphone A (1), microphone B (2) is connected on support (9), arrange in H type, described preamplifier A (3), preamplifier B (4) is connected to signal acquisition module (5), described signal acquisition module (5), memory module (8), display (7) is connected in Sound sources identification and localization module (6), microphone A (1) and microphone B (2) periphery are all enclosed with electromic thermostat (10).

2. one according to claim 1 portable two microphones Sound sources identification and localization device, it is characterized in that, described microphone A (1) is connected with preamplifier A (3) by screw thread, and microphone B (2) is connected with preamplifier B (4) by screw thread.

3. one according to claim 1 portable two microphones Sound sources identification and localization device, is characterized in that, the directive property of described microphone A (1) and described microphone B (2) is omni-directional.

4. one according to claim 1 portable two microphones Sound sources identification and localization device, it is characterized in that, described support (9) is " Н " type, wherein two is circular groove, described microphone A (1) and described microphone B (2) embed support (9) both sides respectively, and keeping parallelism is arranged.

5. one according to claim 1 portable two microphones Sound sources identification and localization device, is characterized in that, long 5 centimetres of horizontal edge in the middle of described " Н " type support (9).

6. one according to claim 1 portable two microphones Sound sources identification and localization device, is characterized in that, described preamplifier A (3) and described preamplifier B (4) are for built-in.

7. one according to claim 1 portable two microphones Sound sources identification and localization device, is characterized in that, described display (7) is liquid crystal display (7).

8. one according to claim 1 portable two microphones Sound sources identification and localization device, it is characterized in that, described memory module (8) stores the institute's sound source obtained by Sound sources identification and localization module (6), and described display (7) shows amplitude and the phase place of institute's sound source by polar diagram.