CN102760461B - A kind of audio-frequence player device of Automatic adjusument volume and method - Google Patents

Abstract

The invention discloses a kind of audio-frequence player device and method of Automatic adjusument volume, it estimates the time delay of back coupling sound by cross correlation algorithm, and using the reference input of the time delay of decoded audio signal as self-adaptive processing, the signal of the voice signal detected as self-adaptive processing is inputted by noise detection module simultaneously.Self-adaptive processing adopts FIR filter, and uses NLMS adaptive filter algorithm.Through self-adaptive processing, suppress the back coupling of normal play, from noise transducer, extract true noise signal, after noise level being weighted to moving average process, regulating the broadcast sound volume size of playback equipment.The present invention is applicable to the more violent occasion of ambient noise change.

Description

A kind of audio-frequence player device of Automatic adjusument volume and method

Technical field

The present invention relates in a kind of audio-frequence player device the environment noise test technology and the adaptive gain regulation technology that suppress to play sound, particularly relate to a kind of can the suppression or the back coupling of broadcasting sound in stress release treatment transducer, environmentally noise size adaptation regulates audio-frequence player device and the method for volume.

Background technology

At the occasion playing audio signal that ambient noise change is more violent, the energy that there will be ambient noise and normal play sound does not mate, and causes audience's auditory fatigue, the phenomenons such as the sense of hearing is unintelligible.If ambient noise is excessive, signal to noise ratio is too low, and audience need pick up one's ears and just can not hear, and causes auditory fatigue; If ambient noise is lower, excessive to play sound and can ear damage hearing.

Realize the detection technique that one of key technology of Gain Automatic control is environmental noise power.Sound wave is propagated in atmosphere, not only comprises ambient noise in noise measuring transducer, also comprises normal play audio signal, only has and is fallen by the echo suppressing of the normal play audio frequency in transducer, just can obtain real environmental noise power.

A kind of existing method, by the special mounting of ambient noise sensor, avoids pickup to play sound, the then noise energy of calculating noise transducer, with this state modulator player gain.In practical application, be difficult to meet this requirement, the broadcasting sound of even noise transducer pickup sometimes all can be better than ambient noise.Move back and ask secondary, another kind of application is in broadcasting interval, when namely broadcasting suspends, and detection noise, and noise level controls foundation as gain in proper order.This method is not suitable for again continuous continual broadcasting occasion.

Perfected process exports noise transducer to carry out signal transacting, the broadcasting sound in restraint speckle transducer pickup, and this not only needs to design suitable algorithm, also needs the hardware computation such as DSP or flush bonding processor platform.There are some patents to relate to describing of this respect at present, but do not have public algorithm and implementation method.

Summary of the invention

The present invention mainly solves the technical problem that being difficult to existing for prior art accurately obtains size of ambient noise, provides a kind of and filtering can play sound, the audio-frequence player device accurately obtaining the Automatic adjusument volume of size of ambient noise and method.

The present invention is directed to that above-mentioned technical problem mainly solved by following technical proposals: a kind of audio-frequence player device of Automatic adjusument volume, comprising:

Processor module, for controlling whole equipment;

Noise detection module, sends to processor module for testing environment noise;

Power amplifier module, amplifies for the play signal exported by processor module;

Loudspeaker, play for the play signal after being amplified by power amplifier module;

Audio interface block, described Audio interface block is connection handling device module, power amplifier module and noise detection module simultaneously, the play signal of processor module, to processor module, is passed to power amplifier module by signal transmission noise detection module detected simultaneously.

As preferably, the audio-frequence player device of Automatic adjusument volume also comprises wireless network module, and described wireless network module is connected with described processor module.

As preferably, described processor module comprises USB interface and SD card interface.

As preferably, the audio-frequence player device of Automatic adjusument volume also comprises analogue audio frequency module, and described analogue audio frequency module is connected with processor module by Audio interface block.

Playing the source of sound of sound can be stored in the audio file in the memory device that is connected with processor module by USB interface or SD card interface, also can be the network audio Streaming Media inputted by wireless network module, can also be the local analogue audio frequency inputted by analogue audio frequency module.

An audio frequency playing method for Automatic adjusument volume, comprises the following steps:

Step one, play audio file from source of sound;

Step 2, testing environment noise, judge the size of ambient noise;

Step 3, environmentally noise size adjustment play the volume of sound.

As preferably, in described step 2, testing environment noise is specially:

A, estimate to feedback the time delay of sound by cross correlation algorithm;

B, time delay preliminary treatment is carried out to reference input;

Back coupling sound in C, the voice signal that detects with NLMS adaptive filter algorithm restraint speckle transducer, thus extract noise signal, by the later reference-input signal of time delay preliminary treatment by FIR sef-adapting filter;

Reference-input signal is the audio signal exported of decoding from the audio file that SD card, USB flash disk etc. store, or the audio signal of output of also being decoded by network reception.

As preferably, in described step 3, the volume that environmentally noise size adjustment plays sound is specially: be first weighted moving average process to noise level, then according to auditory properties, gain and noise level regulate by arc tangent curve.

As preferably, described weighted connections is:

${w^{\′}}_{\mathrm{noise}}\left(n\right)=\frac{1}{35}[-3w_{\mathrm{noise}}(n-2)+12w_{\mathrm{noise}}(n-1)+17w_{\mathrm{noise}}\left(n\right)+12w_{\mathrm{noise}}(n+1)-3w_{\mathrm{noise}}(n+2)]$

W _noisethe environmentally noise level that calculates of noise signal η (n), w ' _noiseit is the later noise level of weighting.

As preferably, described gain-adjusted curve is:

A＝f(w′ _noise)＝tan ^-1(w′ _noise)

A is system gain.

The present invention adopts the broadcasting sound in auto-adaptive filtering technique restraint speckle transducer, and audio signal player inner decoding exported is as reference input, and the input of noise transducer inputs as signal.Conventional adaptive echo elimination algorithm has least-mean-square error algorithm LMS, Normalized least mean squares NLMS, RLS RLS etc.The feature of LMS adaptive algorithm is that complexity is low, and robustness is good, and steady-state error and convergence rate can not be simultaneously optimum.NLMS is a kind of important New variable step-size LMS, and it is on succession LMS algorithm advantage basis, and when error is larger, step-length is comparatively large, can convergence speedup speed; When error is stablized, step-length is less, can reduce steady-state error.RLS convergence rate is faster than LMS, and steady-state error is also less, but computation complexity is more much larger than LMS and NLMS algorithm.

In NLMS algorithm, sef-adapting filter adopts FIR type digital filter, and filter order is a key parameter, and it is relevant to the sound field environment residing for player, and its size can from tens rank to thousands of rank not etc.Wave-wave device order is on the low sidely fixed, and inhibition reduces; If order is too highly fixed, computation complexity can be increased in vain again.Sound field environment due to practical application has very large uncertainty, brings difficulty to the setting of sef-adapting filter order.

In order to reduce computation complexity, first the present invention estimates the time delay of playing sound by cross correlation algorithm, then time delay preliminary treatment is carried out to decoding output audio signal (i.e. reference input), so just can adopt 128 fixing rank FIR sef-adapting filters, reduce computational complexity, last again by NLMS adaptive filter algorithm extraction noise signal, add convergence rate, decrease steady-state error.

After obtaining noise level estimation, suitable gain-adjusted control algolithm must be designed.Gain-adjusted algorithm and automatic growth control (AGC) algorithm similar, and must regulate between stationarity compromise in response speed.The present invention, is first weighted moving average process to noise level, reduces the violent degree of change in gain, and then according to auditory properties, gain and noise level regulate by arc tangent curve.

The present invention can be applicable to the occasions such as the outdoor digital audio broadcasting such as bus station, railway station, subway platform, bus-stop, and the occasions such as indoor digital audio frequency broadcasting such as bar, coffee shop, background music, application the present invention can obtain constant signal to noise ratio, reaches best audio frequency sense of hearing and audio frequency definition.

The substantial effect that the present invention brings is, (1) proposes a kind ofly to follow the integration algorithm that ambient noise level regulates player gain automatically, meets various broadcasting Automatic level control needs with the self-reacting occasion demand of ambient noise.(2) extracting noise level algorithm adopts cross correlation algorithm to estimate the time delay of audio plays, and then adopts NLMS adaptive filter algorithm to eliminate the back coupling of audio plays; (3) adopt weighting moving average algorithm to reduce the fluctuation of gain-adjusted, adopt the relation between arctan function adjustment gain and noise level.

Accompanying drawing explanation

Fig. 1 is a kind of structural representation of the present invention;

Fig. 2 is a kind of algorithm block diagram of the present invention;

Fig. 3 is the block diagram of a kind of auto-adaptive fir filter of the present invention;

Fig. 4 is a kind of automatic growth control flow chart of the present invention;

Fig. 5 is the graph of a relation between a kind of noise level of the present invention and system gain;

In figure: 101, processor module, 102, SD card or USB flash disk, 103, wireless network module, 104, analogue audio frequency module, 105, Audio interface block, 106, power amplifier module, 107, loudspeaker, 108, user interface, 109, noise detection module.

Embodiment

Below by embodiment, and by reference to the accompanying drawings, technical scheme of the present invention is described in further detail.

Embodiment: the audio-frequence player device of a kind of Automatic adjusument volume of the present embodiment, as shown in Figure 1, processor module 101, Audio interface block 105, power amplifier module 106, loudspeaker 107, noise detection module 109, analogue audio frequency module 104 and wireless network module 103 is comprised.Audio interface block 105 is connection handling device module 101, power amplifier module 106, noise detection module 109 and analogue audio frequency module 104 simultaneously, the play signal of processor module 101, to processor module 101, is passed to power amplifier module 106 by signal transmission noise detection module 109 detected simultaneously.Power amplifier module 106 is play by loudspeaker 107 after play signal being amplified.Processor module 101 includes USB interface and SD card interface.Playing the source of sound of sound can be stored in the audio file in the USB flash disk or SD card 102 that are connected with processor module by USB interface or SD card interface, also can be the network audio Streaming Media inputted by wireless network module 103, can also be the local analogue audio frequency inputted by analogue audio frequency module 104.

Processor module 101 is also connected with user interface 108, and user can carry out optimum configurations by user interface 108 pairs of equipment.

The audio signal that noise detection module 109 detects, except ambient noise, also comprises the back coupling of playing sound.

The audio frequency playing method of a kind of Automatic adjusument volume of the present embodiment, comprises the following steps:

Step one, play audio file from source of sound;

Step 2, testing environment noise, judge the size of ambient noise;

Step 3, environmentally noise size adjustment play the volume of sound.

Testing environment noise is specially:

A, estimate to feedback the time delay of sound by cross correlation algorithm;

B, time delay preliminary treatment is carried out to decoded audio signal (reference input);

Back coupling sound in C, the voice signal that detects with NLMS adaptive filter algorithm restraint speckle transducer, thus extract noise signal.

Environmentally the volume of noise size adjustment broadcasting sound is specially: be first weighted moving average process to noise level, and then according to auditory properties, gain and noise level regulate by arc tangent curve.

The present embodiment utilizes cross correlation algorithm and adaptive filter algorithm to suppress to play back coupling or the echo of sound.X (n) is normal play audio signal, f (n) is the audio signal that noise detection module detects, echo-signal y (n)=T [x (n)] that ambient noise signal η (n) produces through space with broadcasting sound is comprised, namely in f (n)

f(n)＝y(n)+η(n)

As shown in Figure 2, the cross-correlation r of f (n) and x (n) is first calculated _xf(n, n+ τ), because x (n) and y (n)=T [x (n)] has similitude, namely the peak point of cross-correlation function is back coupling time delay, analyzes cross correlation results, obtains the echo time delay K of x (n).X (n) is through time delay K sampled point x _d(n)=x (n-K).X _dn () enters sef-adapting filter and carries out echo suppressing together with f (n), obtain ambient noise signal η (n).

Fig. 3 is the implementation structure of auto-adaptive fir filter.FIR filter adopts transversal type structure, and its unit impulse response H (n) is:

H(n)＝{h _n1，h _n2,..,h _nM}

X _dn () is through the vector of the input time delay after cross correlation process:

X _D(n)＝{x _D(n),x _D(n-1),...,x _D(n-M+1)}

the echo estimation of audio plays:

$\hat{y}\left(n\right)=\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{h}_i\left(n\right)\·x_D(n-i+1)=H{\left(n\right)}^T\·X_D\left(n\right)$

The target function of suppression echo algorithm is:

J＝min{||η(n)|| ²}， $\mathrm{\η}\left(n\right)=f\left(n\right)-\hat{y}\left(n\right)$

The present invention adopts variable step NLMS algorithm to make objective optimization, and the recurrence formula of sef-adapting filter unit impulse response H (n) is:

$H(n+1)=H\left(n\right)+\mathrm{\μ}\·\frac{\∂{\left|\right|\mathrm{\η}\left(n\right)\left|\right|}^2}{\∂H}=H\left(n\right)-2\mathrm{\μ}\left(n\right)\·\mathrm{\η}\left(n\right)\·X_D\left(n\right)$

$\mathrm{\&mu;}(n+1)=\left\{\begin{array}{c}{\mathrm{\&mu;}}_{\min },\mathrm{\&mu;}<{\mathrm{\&mu;}}_{\min }\\ \mathrm{\&alpha;}\&CenterDot;\mathrm{\&mu;}\left(n\right)+\mathrm{\&beta;}\&CenterDot;{\left|r_{\mathrm{\&eta;x}}\left(n\right)\right|}^2\\ {\mathrm{\&mu;}}_{\max },\mathrm{\&mu;}>{\mathrm{\&mu;}}_{\max }\end{array}\right.$

Wherein, α is step-length gene, and interval is usually between 0.95 ~ 0.999, and the larger steady-state error of this value is larger.β is the instantaneous energy of noise and audio signal cross-correlation function.When noise and audio signal divide deviation, cross-correlation instantaneous energy is large, large to step-length contribution, can convergence speedup speed.β interval between 0 ~ 1, usually close to 0, typically value 10 ^-4~ 10 ^-6, convergence stability and this value are inversely proportional to, and convergence rate is directly proportional to this value.

Fig. 4 is automatic growth control AGC flow process, w _noisebe the environmentally noise level (unit: decibel) that calculates of noise signal η (n), carry out 5 weighting moving averages to noise level, weighted connections is:

${w^{\&prime;}}_{\mathrm{noise}}\left(n\right)=\frac{1}{35}[-3w_{\mathrm{noise}}(n-2)+12w_{\mathrm{noise}}(n-1)+17w_{\mathrm{noise}}\left(n\right)+12w_{\mathrm{noise}}(n+1)-3w_{\mathrm{noise}}(n+2)]$

Through transfer function, the noise gain after level and smooth is converted to the yield value of audio plays.

Fig. 5 is the relation curve between noise level and system gain.W ' _noisebe the later noise level of weighting moving average, A is system gain, is gain-adjusted curve,

A＝f(w′ _noise)＝tan ^-1(w′ _noise)

Coordinate (0,0) place is reference noise level and reference gain value.Usually determining reference noise level is 60db (open air), and reference gain value is 0db.Gain-adjusted scope allows system gain A to increase 12dB on the basis of datum mark, or reduce 12dB.Dynamic adjustments scope is 24dB.

The control system that the present embodiment application flush bonding processor is formed realizes said process.

Specific embodiment described herein is only to the explanation for example of the present invention's spirit.Those skilled in the art can make various amendment or supplement or adopt similar mode to substitute to described specific embodiment, but can't depart from spirit of the present invention or surmount the scope that appended claims defines.

Although more employ noise measuring herein, suppress the terms such as echo, do not get rid of the possibility using other term.These terms are used to be only used to describe and explain essence of the present invention more easily; The restriction that they are construed to any one additional is all contrary with spirit of the present invention.

Claims (3)

1. an audio frequency playing method for Automatic adjusument volume, is characterized in that comprising the following steps:

Step one, play audio file from source of sound;

Step 2, testing environment noise, judge the size of ambient noise;

Step 3, environmentally noise size adjustment play the volume of sound;

In described step 2, testing environment noise is specially:

The cross-correlation of A, the audio signal detected by cross correlation algorithm calculating noise detection module and normal play audio signal, estimates to feedback the time delay of sound;

B, time delay preliminary treatment is carried out to reference input;

Back coupling sound in C, the voice signal that detects with NLMS adaptive filter algorithm restraint speckle transducer, thus extract noise signal;

In described step 3, environmentally the volume of noise size adjustment broadcasting sound is specially: be first weighted moving average process to noise level, and then according to auditory properties, gain and noise level regulate by arc tangent curve.

2. the audio frequency playing method of a kind of Automatic adjusument volume according to claim 1, it is characterized in that, described weighted connections is:

${w^{\&prime;}}_{\mathrm{noise}}\left(n\right)=\frac{1}{35}[-3w_{\mathrm{noise}}(n-2)+12w_{\mathrm{noise}}(n-1)+17w_{\mathrm{noise}}\left(n\right)+12w_{\mathrm{noise}}(n+1)-3w_{\mathrm{noise}}(n+2)]$

W _noisethe environmentally noise level that calculates of noise signal η (n), w' _noiseit is the later noise level of weighting.

3. the audio frequency playing method of a kind of Automatic adjusument volume according to claim 1, is characterized in that, described gain-adjusted curve is:

A＝f(w' _noise)＝tan ^-1(w' _noise)

A is system gain.