CN104464752A - Sound feedback detection method and device - Google Patents

Abstract

The invention discloses a sound feedback detection method and device. The method comprises the steps that time-frequency conversion is conducted on received time-domain signals to obtain frequency-domain signals; the power sum value of a plurality of adjacent points on the left and right of power peak values and the average power value of the frequency-domain signals are calculated; decision values are determined according to the power sum value and the average power value; a corresponding preset first threshold value is determined according to the frequency band range which the power peak value frequency falls into, and if the decision values are larger than the preset first threshold value, it is determined that the decision values are decision values to be counted corresponding to the frequency band range; the number of the decision values to be counted corresponding to the frequency band range and the repetition time duration of the power peak values falling into the frequency band range in the preset time are counted; if the number is larger a preset second threshold value or the repetition time duration is larger than a preset third threshold value, it is determined that sound feedback is in process. According to the technical scheme, the accuracy and reliability of sound feedback detection can be improved.

Description

A kind of acoustic feedback detection method and device

Technical field

The present invention relates to wireless communication field, particularly a kind of acoustic feedback detection method and device.

Background technology

Acoustic feedback refers to that the sound sent from loudspeaker entered feedback path and again feeds back to microphone, by the phenomenon of microphone pickup; Due to the existence of acoustic feedback, make final sound field Frequency Response bad, extremely ear-piercing uttering long and high-pitched sounds can be produced time serious, tonequality severe exacerbation.

In order to suppress acoustic feedback, prior art adopts following technical scheme " for detecting the apparatus and method of acoustic feedback ", this device comprises: the first level detection parts, is configured to the signal level detecting the voice signal that the position from the voice signal system being connected to microphone and loudspeaker obtains; First extracting parts, is configured to extract from the voice signal being detected signal level to have for the signal in the frequency band in each predetermined broadband at least one predetermined center frequency; Second electrical level detection part, is configured to the signal level of the signal detected in each frequency band, and described signal is extracted by the first extracting parts; And decision means, be configured to judge whether acoustic feedback to occur based on according to the determined threshold value of waveform of the signal level detected by the first level detection parts and each signal level detected by second electrical level detection part.

Above-mentioned prior art realizes the first extracting parts by multiple bandpass filter, and bandpass filter is subject to the restriction of centre frequency and bandwidth, cannot cover Whole frequency band, easily occurs undetected problem; In addition, by judging that the mode whether signal level waveform has a sine wave period determines whether acoustic feedback to occur, the easy like this problem occurring flase drop.

Summary of the invention

In view of this, fundamental purpose of the present invention is to provide a kind of acoustic feedback detection method and device, to solve the problem of the undetected and flase drop caused due to broadband filter and signal waveform detection acoustic feedback, improves acoustic feedback and accuracy and reliability detected.

In first aspect present invention, provide a kind of acoustic feedback detection method, comprising:

Time-frequency conversion is carried out to the time-domain signal received and obtains corresponding frequency-region signal;

According to described frequency-region signal determination power peak, calculate the adjacent power of multiple point in described power peak left and right and the average power content of value and described frequency signal;

According to described power and value and described average power content determination decision value;

Determine corresponding preset first threshold according to the band limits that described power peak frequency falls into, if described decision value is greater than described preset first threshold, then determine that described decision value is the decision value to be counted that described band limits is corresponding;

The number of statistics decision value described to be counted that described band limits is corresponding in preset time, and the power peak fallen in preset time in described band limits repeats duration; If described number is greater than preset Second Threshold or described repetition duration when being greater than the 3rd preset threshold value, then determine acoustic feedback is occurring.

Preferably, described determine acoustic feedback to occur after, described method also comprises:

Be decreased to the mode of desired value according to gain coefficient gradually, attenuation processing carried out to described time-domain signal and obtains the signal after acoustic feedback suppression.

Preferably, described method also comprises:

Identify whether the amplitude of described time-domain signal is less than the 4th preset threshold value;

Signal amplitude being less than to the 4th preset threshold value obtains the time-domain signal after small-signal decay as attenuation processing;

The then described mode being decreased to desired value according to gain coefficient gradually, attenuation processing is carried out to described time-domain signal and obtains the signal after acoustic feedback suppression, be specially:

Be decreased to the mode of desired value according to gain coefficient gradually, attenuation processing carried out to the time-domain signal after described small-signal decay and obtains the signal after acoustic feedback suppression.

Preferably, if described number be less than or equal to preset Second Threshold and described repetition duration is less than or equal to the 3rd preset threshold value time, then determine acoustic feedback does not occur,

Then described method also comprises:

Calculate the maximum likelihood ratio of speech frame according to described time-domain signal and noise power-value, if described maximum likelihood ratio is greater than preset voice threshold value, determine that described time-domain signal is voice signal;

Increase to the mode of 0dB according to described gain coefficient gradually, gain process is carried out to described voice signal and obtains the signal after gain process.

Preferably, to be describedly specially according to described power and value and described average power content determination decision value:

Calculate the ratio of described power and value and described average power content, determine that described ratio is decision value.

Preferably, according to described frequency-region signal determination power peak, calculate the adjacent power of multiple point in described power peak left and right and the average power content of value and described frequency signal, specifically comprise:

According to formula the adjacent multiple point in rated output peak value left and right and value, wherein, X max (0) is power peak; When j is not equal to 0, Xmax (j) is expressed as the adjacent multiple points in power peak left and right; K is more than or equal to 1;

According to formula $\mathrm{Pav}=[\underset{j=0}{\overset{N/2-1}{\mathrm{\Σ}}}\mathrm{Xj}-\underset{j=-k}{\overset{k}{\mathrm{\Σ}}}X\max \left(j\right)]/(\frac{N}{2}-2k-1)$ Calculate average power content, wherein, represent all performance numbers of described frequency signal power spectrum and value.

In second aspect present invention, provide a kind of acoustic feedback pick-up unit, comprising:

Time-frequency conversion unit, obtains corresponding frequency-region signal for carrying out time-frequency conversion to the time-domain signal received;

Computing unit, for according to described frequency-region signal determination power peak, calculates the adjacent power of multiple point in described power peak left and right and the average power content of value and described frequency signal;

Decision value determining unit, for according to described power and value and described average power content determination decision value;

Decision value determining unit to be counted, band limits for falling into according to described power peak frequency determines corresponding preset first threshold, if described decision value is greater than described preset first threshold, then determine that described decision value is the decision value to be counted that described band limits is corresponding;

Acoustic feedback determining unit, for adding up the number of decision value described to be counted corresponding to described band limits in preset time, and the power peak fallen in preset time in described band limits repeats duration; If described number is greater than preset Second Threshold or described repetition duration when being greater than the 3rd preset threshold value, then determine acoustic feedback is occurring.

Preferably, described device also comprises:

Suppress unit, for being decreased to the mode of desired value gradually according to gain coefficient, attenuation processing being carried out to described time-domain signal and obtains the signal after acoustic feedback suppression.

Preferably, described device also comprises:

Recognition unit, for identifying whether the amplitude of described time-domain signal is less than the 4th preset threshold value;

Small-signal attenuation units, obtains the time-domain signal after small-signal decay for signal amplitude being less than to the 4th preset threshold value as attenuation processing;

Then described suppression unit, specifically for being decreased to the mode of desired value gradually according to gain coefficient, carrying out attenuation processing to the time-domain signal after described small-signal decay and obtains the signal after acoustic feedback suppression.

Preferably, described device also comprises:

Voice determining unit, for calculating the maximum likelihood ratio of speech frame according to described time-domain signal and noise power-value, if described maximum likelihood ratio is greater than preset voice threshold value, determines that described time-domain signal is voice signal;

Speech gain processing unit, for increasing to the mode of 0dB gradually according to described gain coefficient, carries out gain process to described voice signal and obtains the signal after gain process.

Preferably, described decision value determining unit, specifically for calculating the ratio of described power and value and described average power content, determines that described ratio is decision value.

Preferably, described computing unit, specifically comprises:

With value computation subunit, for according to formula the adjacent multiple point in rated output peak value left and right and value, wherein, X max (0) is power peak; When j is not equal to 0, Xmax (j) is expressed as the adjacent multiple points in power peak left and right; K is more than or equal to 1;

Average power content computation subunit, for

According to formula $\mathrm{Pav}=[\underset{j=0}{\overset{N/2-1}{\mathrm{\Σ}}}\mathrm{Xj}-\underset{j=-k}{\overset{k}{\mathrm{\Σ}}}X\max \left(j\right)]/(\frac{N}{2}-2k-1)$ Calculate average power content, wherein, represent all performance numbers of described frequency signal power spectrum and value.

In third aspect present invention, provide a kind of acoustic feedback pick-up unit, it is characterized in that, comprise at least one processor, at least one network interface or other communication interface, storer, and at least one communication bus; Described storer is used for stored program instruction, and described processor is used for performing following steps according to described programmed instruction:

Time-frequency conversion is carried out to the time-domain signal received and obtains corresponding frequency-region signal;

According to described frequency-region signal determination power peak, calculate the adjacent power of multiple point in described power peak left and right and the average power content of value and described frequency signal;

According to described power and value and described average power content determination decision value;

Determine corresponding preset first threshold according to the band limits that described power peak frequency falls into, if described decision value is greater than described preset first threshold, then determine that described decision value is the decision value to be counted that described band limits is corresponding;

The number of statistics decision value described to be counted that described band limits is corresponding in preset time, and the power peak fallen in preset time in described band limits repeats duration; If described number is greater than preset Second Threshold or described repetition duration when being greater than the 3rd preset threshold value, then determine acoustic feedback is occurring.As can be seen from technique scheme, the present invention proposes to identify whether acoustic feedback occurs from the frequency characteristic of acoustic feedback signal, mainly utilize power peak and average peak determination decision value, whether acoustic feedback is there is according to the size of this decision value and the duration gauge signal of power peak, like this, neither need the filtering process of bandpass filter, do not need to go sense cycle by signal waveform yet, just can carry out detection signal according to the frequency characteristic of signal and whether acoustic feedback occur, undetected and problem that is flase drop can be avoided like this, improve accuracy and the reliability of acoustic feedback detection.

Accompanying drawing explanation

In order to be illustrated more clearly in the embodiment of the present invention or technical scheme of the prior art, be briefly described to the accompanying drawing used required in embodiment or description of the prior art below, apparently, accompanying drawing in the following describes is only some embodiments of the present invention, for those of ordinary skill in the art, under the prerequisite not paying creative work, other accompanying drawing can also be obtained according to these accompanying drawings.

Fig. 1 is the process flow diagram of acoustic feedback detection method embodiment 1 of the present invention;

Fig. 2 is the process flow diagram of acoustic feedback detection method embodiment 2 of the present invention;

Fig. 3 is the control chart of acoustic feedback automatic gain of the present invention and speed;

Fig. 4 is the process flow diagram of acoustic feedback detection method embodiment 3 of the present invention;

Fig. 5 is the schematic diagram of acoustic feedback pick-up unit embodiment 1 of the present invention;

Fig. 6 is the schematic diagram of acoustic feedback pick-up unit embodiment 2 of the present invention;

Fig. 7 is the schematic diagram of acoustic feedback pick-up unit embodiment 3 of the present invention;

Fig. 8 is that acoustic feedback pick-up unit hardware of the present invention forms schematic diagram.

Embodiment

For enabling above-mentioned purpose of the present invention, feature and advantage become apparent more, are described in further detail the embodiment of the present invention below in conjunction with the drawings and specific embodiments.

Embodiment 1

See Fig. 1, the process flow diagram of the acoustic feedback detection method embodiment 1 of the present invention illustrated, the method can comprise:

S101: time-frequency conversion is carried out to the time-domain signal received and obtains corresponding frequency-region signal.

Technical solution of the present invention is mainly used in professional wireless communication audio equipment, as mobile phone, intercom etc., to lay down a definition explanation below for intercom to content of the present invention.

Intercom receives the signal that other intercoms send, and first needs to do sub-frame processing to signal, and all process and program are all unit based on frame; Then needing each frame signal to receiving to carry out windowing process, finally using FFT to transform to frequency domain; Like this, frequency-region signal corresponding to each frame time-domain signal can be obtained.

S102: according to described frequency-region signal determination power peak, calculates the adjacent power of multiple point in described power peak left and right and the average power content of value and described frequency signal.

In the embodiment of the present invention, this step can realize in the following manner, specifically comprises:

According to formula the adjacent multiple point in rated output peak value left and right and value, wherein, X max (0) is the power peak in power spectrum; When j is not equal to 0, X max (j) is expressed as the adjacent multiple points in power peak left and right; K is more than or equal to 1;

According to formula $\mathrm{Pav}=[\underset{j=0}{\overset{N/2-1}{\mathrm{\Σ}}}\mathrm{Xj}-\underset{j=-k}{\overset{k}{\mathrm{\Σ}}}X\max \left(j\right)]/(\frac{N}{2}-2k-1)$ Calculate average power content, wherein, represent power spectrum all performance numbers and value.

When specific implementation, can also according to formula calculate average power content.

When specific implementation, different equipment is due to the difference of himself frequency characteristic, and the difference of counting out that calculative power peak left and right is adjacent, that is, is applied to different equipment, and in formula, the large I of k value is different above.

S103: according to described power and value and described average power content determination decision value.

When specific implementation, this step can realize in the following manner:

Calculate the ratio of described power and value and described average power content, determine that described ratio is decision value.

S104: determine corresponding preset first threshold according to the band limits that described power peak frequency falls into, if described decision value is greater than described preset first threshold, then determine that described decision value is the decision value to be counted that described band limits is corresponding.

When specific implementation, equipment is in advance according to self Frequency Response determination acoustic feedback generation frequency range and corresponding first threshold, it is noted herein that full frequency band must be divided into multiple frequency range to determine easily to occur each frequency range of acoustic feedback and corresponding first threshold by equipment.

Such as: full frequency band is divided into following three parts according to self Frequency Response by certain intercom, and determine the first threshold that each frequency range is corresponding, specific as follows:

Part I: band limits (2400 ~ 3700Hz), decision value R> first threshold Thresh1, peak follow duration Tr>200ms.

Part II: band limits (1200 ~ 1800Hz), decision value R> first threshold Thresh2, peak follow duration Tr>240ms.

Part III: band limits (other frequency ranges), decision value R> first threshold Thresh3, peak follow duration Tr>300ms.

Below for intercom situation above, explanation is further explained to this step S104.

The power peak that S102 determines above its concrete frequency known, its band limits fallen into is determined according to this power peak frequency, determine corresponding first threshold according to fallen into band limits again, finally determine whether decision value is the decision value to be counted that certain band limits is corresponding.

Suppose that this power peak frequency is 2600Hz, it falls into the band limits (2400 ~ 3700Hz) of Part I, the first threshold of its correspondence is Thresh1, now, judge whether the decision value that S103 determines is greater than Thresh1, if be greater than, then determine that this decision value is the decision value to be counted that Part I frequency range 2400 ~ 3700Hz is corresponding.

Suppose that this power peak is 1600Hz, it falls into the band limits (1200 ~ 1800Hz) of Part II, the first threshold of its correspondence is Thresh2, now, judge whether the decision value that S103 determines is greater than Thresh2, if be greater than, then determine that this decision value is the decision value to be counted that Part II frequency range 2400 ~ 3700Hz is corresponding.

Suppose that this power peak is 1000Hz, it falls into the band limits (other frequency ranges) of Part III, the first threshold of its correspondence is Thresh3, now, judge whether the decision value that S103 determines is greater than Thresh3, if be greater than, then determine that this decision value is the decision value to be counted that Part III band limits is corresponding.

S105: the number of adding up decision value to be counted corresponding to described band limits in preset time, and the power peak fallen in preset time in described band limits repeats duration; If described number is greater than preset Second Threshold or described repetition duration when being greater than the 3rd preset threshold value, then determine acoustic feedback is occurring.

When specific implementation, while equipment pre-sets first threshold corresponding to each frequency range, also need to arrange Second Threshold corresponding to each frequency range and the 3rd threshold value.When this step specific implementation, the total number of decision value to be counted that the band limits needing statistics S104 to determine is corresponding, also to add up the repetition duration of the power peak fallen in this band limits simultaneously, the last preset Second Threshold corresponding for this frequency range and the 3rd threshold value are done to judge further, if described number is greater than preset Second Threshold or described repetition duration when being greater than the 3rd preset threshold value, then determine acoustic feedback is occurring.

Because the present invention is the important evidence that the band limits fallen into by power peak frequency is detected as acoustic feedback, the division of band limits here take Whole frequency band as benchmark, therefore it is without the need to bandpass filter detection signal, also would not occur because the centre frequency of bandpass filter and bandwidth arrange the undetected problem that size causes.The present invention is by the number of crest frequency repetition duration with the decision value satisfied condition in addition, these two values are all whether carry out gauge signal from the angle of the frequency speciality of acoustic feedback be acoustic feedback signal, without the need to going to judge signal period property from the angle of signal waveform, and the computation process of these two values is simply not easily made mistakes, thus avoid the problem utilizing waveform to judge to cause flase drop.

Embodiment 2

See Fig. 2, the process flow diagram of the acoustic feedback detection method embodiment 2 of the present invention illustrated, the method mainly increases the processing mode suppressing acoustic feedback on foregoing embodiments 1 basis, to reach the object suppressing acoustic feedback to improve speech quality.The method specifically comprises:

S201: time-frequency conversion is carried out to the time-domain signal received and obtains corresponding frequency-region signal;

S202: according to described frequency-region signal determination power peak, calculates the adjacent power of multiple point in described power peak left and right and the average power content of value and described frequency signal;

S203: according to described power and value and described average power content determination decision value;

S204: determine corresponding preset first threshold according to the band limits that described power peak frequency falls into, if described decision value is greater than described preset first threshold, then determine that described decision value is the decision value to be counted that described band limits is corresponding;

S205: the number of adding up decision value described to be counted corresponding to described band limits in preset time, and the power peak fallen in preset time in described band limits repeats duration; If described number is greater than preset Second Threshold or described repetition duration when being greater than the 3rd preset threshold value, then determine acoustic feedback is occurring.

S201-S205 is identical with the S101-S105 described in foregoing embodiments 1, does not repeat them here.

S206, is decreased to the mode of desired value gradually according to gain coefficient, carry out attenuation processing obtain the signal after acoustic feedback suppression to the described time-domain signal received.

Wherein, desired value refers to targeted security critical gain, and this desired value can be the gain that equipment tests that the just guarantee equipment obtained does not produce acoustic feedback in advance.

During specific implementation, equipment can the fall off rate Vdown dB/s of preset gain coefficient, according to gain coefficient from 0dB, can change according to fall off rate Vdown, until gain coefficient reaches desired value.The value of different its Vdown of equipment is not identical yet, but Vdown must be less than zero.

When S206 is for signal generation acoustic feedback above, the process suppressing acoustic feedback is done to signal, but in actual applications, also need to pay close attention to when acoustic feedback disappears, need to ensure that voice signal recovers preferably, avoid the problem that sound suddenlys change, tonequality is poor, the voice experience of user can be improved like this.

For solving the above-mentioned problem about voice, the present invention also provides preferred version, and this preferred version specifically adds the method for speech gain process on the basis of above-described embodiment 2, specifically comprises:

If described number be less than or equal to preset Second Threshold and described repetition duration is less than or equal to the 3rd preset threshold value time, then determine acoustic feedback does not occur, then said method also comprises:

Calculate the maximum likelihood ratio of speech frame according to described time-domain signal and noise power-value, if described maximum likelihood ratio is greater than preset voice threshold value, determine that described time-domain signal is voice signal;

Increase to the mode of 0dB according to described gain coefficient gradually, gain process is carried out to described voice signal and obtains the signal after gain process.

See Fig. 3, the acoustic feedback automatic gain illustrated and speed control chart, in Fig. 3, fall off rate VdowndB/s is the gain fall off rate that acoustic feedback suppresses, rising Vup dB/s is the gain climbing speed that speech gain compensates, and when specific implementation, equipment can pre-set climbing speed Vup and Vdown of gain coefficient, when finding signal generation acoustic feedback, by gain coefficient from 0dB, change, until gain coefficient reaches desired value according to fall off rate Vdown dB/s; According to the mode of gain coefficient change, suppression process is carried out to acoustic feedback signal; Detected when there is no acoustic feedback afterwards, when namely only having pure voice, this needs to carry out gain compensation to voice signal, by gain coefficient from desired value before, change according to according to climbing speed Vup dB/s, until gain coefficient reaches 0dB, the mode increased gradually according to gain coefficient processes signal.

Here it should be noted that, the Vup numerical value possibility that different equipment is arranged is identical also may not be identical, but Vup numerical value must be greater than zero; In order to improve suppression and speech gain effect, Vup absolute value can also be set and be less than Vdown absolute value, can ensure like this to suppress acoustic feedback in the equipment very fast time, and recover voice signal slowly, acoustic feedback can be avoided the impact of user, can ensure that again voice better recover.

This preferred version is when finding that acoustic feedback does not occur signal, making further gain compensation to voice signal makes it suppress the gain after processing to rise to normal level with given pace by acoustic feedback, so just can avoid the problem of voice sudden change, both poor sound quality to a certain extent.

Embodiment 3

In addition, due to voice tail place little noise easily received machine pick up the source as causing acoustic feedback, and then providing occurrence condition for acoustic feedback, in order to reduce acoustic feedback probability of happening, present invention also offers preferred version.The program, specifically on the basis of above-described embodiment 2, adds the noise section in small-signal damped system deamplification, avoids it to cause acoustic feedback.

See Fig. 4, the process flow diagram of the acoustic feedback detection method embodiment 3 of the present invention illustrated, the method mainly increases the processing mode of small-signal decay on foregoing embodiments 2 basis, to reach the noise decrease that declines, avoids the object causing acoustic feedback.The method specifically comprises:

S301: time-frequency conversion is carried out to the time-domain signal received and obtains corresponding frequency-region signal;

S302: according to described frequency-region signal determination power peak, calculates the adjacent power of multiple point in described power peak left and right and the average power content of value and described frequency signal;

S303: according to described power and value and described average power content determination decision value;

S304: determine corresponding preset first threshold according to the band limits that described power peak frequency falls into, if described decision value is greater than described preset first threshold, then determine that described decision value is the decision value to be counted that described band limits is corresponding;

S305: the number of adding up decision value described to be counted corresponding to described band limits in preset time, and the power peak fallen in preset time in described band limits repeats duration; If described number is greater than preset Second Threshold or described repetition duration when being greater than the 3rd preset threshold value, then determine acoustic feedback is occurring.

S301-S305 is identical with the S101-S105 described in foregoing embodiments 1, does not repeat them here.

S306: identify whether the amplitude of described time-domain signal is less than the 4th preset threshold value;

S307: signal amplitude being less than to the 4th preset threshold value obtains the time-domain signal after small-signal decay as attenuation processing;

Concrete, S306 and S307 can be realized by following formula group:

X in formula is the time-domain signal received, y is the time-domain signal after small-signal decay, threshold is preset 4th preset, k is gain situation according to equipment self and sets, above-mentioned formula can be understood as: when the amplitude of time-domain signal be less than preset 4th preset time, according to the formula of the first row in formula group, signal is processed, if the amplitude of time-domain signal be more than or equal to preset 4th preset time, according to the formula of the second row in formula group, namely keep original signal constant.

Here it should be noted that, preset the 4th preset object is that maximum possible reduces attenuation processing to the impact of voice signal, the energy of voice signal is concentrated on threshold value substantially, to reduce the decay to voice signal, only attenuation processing is carried out to small-signal, so just can reduce the quality that acoustic feedback probability of happening ensures voice signal simultaneously to a certain extent.

S308: the mode being decreased to desired value according to gain coefficient gradually, carries out attenuation processing to the time-domain signal after described small-signal decay and obtains the signal after acoustic feedback suppression.

Here it should be noted that, execution sequence requirement not strict between each step S301-S308 of the present embodiment, when specific implementation, can perform according to the order shown in Fig. 3, also while order performs S301-S305, order S306-307 can be performed, finally perform S308, other execution sequences can certainly be adopted to realize the present invention.

By the amplitude of identification signal, this preferred version, judges whether signal is little noise, if little noise is making attenuation processing to this signal, so just little noise can be destroyed, avoid follow-up due to little noise introducing acoustic feedback.

Embodiment 4

Corresponding with above-described embodiment 1-3, present invention also offers acoustic feedback pick-up unit, this device can be applicable in Wireless Telecom Equipment, as intercom, mobile phone etc.

See Fig. 5, the schematic diagram of the acoustic feedback pick-up unit embodiment 1 of the present invention illustrated, this device can comprise:

Time-frequency conversion unit 401, obtains corresponding frequency-region signal for carrying out time-frequency conversion to the time-domain signal received;

Computing unit 402, for according to described frequency-region signal determination power peak, calculates the adjacent power of multiple point in described power peak left and right and the average power content of value and described frequency signal;

Decision value determining unit 403, for according to described power and value and described average power content determination decision value;

Decision value determining unit 404 to be counted, band limits for falling into according to described power peak frequency determines corresponding preset first threshold, if described decision value is greater than described preset first threshold, then determine that described decision value is the decision value to be counted that described band limits is corresponding;

Acoustic feedback determining unit 405, for adding up the number of decision value described to be counted corresponding to described band limits in preset time, and the power peak fallen in preset time in described band limits repeats duration; If described number is greater than preset Second Threshold or described repetition duration when being greater than the 3rd preset threshold value, then determine acoustic feedback is occurring.

Preferably, described decision value determining unit, specifically for calculating the ratio of described power and value and described average power content, determines that described ratio is decision value.

Preferably, computing unit, specifically comprises:

With value computation subunit, for according to formula the adjacent multiple point in rated output peak value left and right and value, wherein, X max (0) is power peak; When j is not equal to 0, X max (j) is expressed as the adjacent multiple points in power peak left and right; K is more than or equal to 1;

Average power content computation subunit, for

According to formula $\mathrm{Pav}=[\underset{j=0}{\overset{N/2-1}{\mathrm{\Σ}}}\mathrm{Xj}-\underset{j=-k}{\overset{k}{\mathrm{\Σ}}}X\max \left(j\right)]/(\frac{N}{2}-2k-1)$ Calculate average power content, wherein, represent all performance numbers of described frequency signal power spectrum and value.

Embodiment 5

See Fig. 6, the schematic diagram of the acoustic feedback pick-up unit embodiment 2 of the present invention illustrated, this device also can comprise on the basis of above-described embodiment 4:

Suppress unit 406, for being decreased to the mode of desired value gradually according to gain coefficient, attenuation processing being carried out to described time-domain signal and obtains the signal after acoustic feedback suppression.

Preferably, described device also comprises:

Voice determining unit, for calculating the maximum likelihood ratio of speech frame according to described time-domain signal and noise power-value, if described maximum likelihood ratio is greater than preset voice threshold value, determines that described time-domain signal is voice signal;

Speech gain processing unit, for increasing to the mode of 0dB gradually according to described gain coefficient, carries out gain process to described voice signal and obtains the signal after gain process.

Embodiment 6

See Fig. 7, the schematic diagram of the acoustic feedback pick-up unit embodiment 3 of the present invention illustrated, this device also can comprise on the basis of above-described embodiment 5:

Recognition unit 407, for identifying whether the amplitude of described time-domain signal is less than the 4th preset threshold value;

Small-signal attenuation units 408, obtains the time-domain signal after small-signal decay for signal amplitude being less than to the 4th preset threshold value as attenuation processing;

Then described suppression unit, specifically for being decreased to the mode of desired value gradually according to gain coefficient, carrying out attenuation processing to the time-domain signal after described small-signal decay and obtains the signal after acoustic feedback suppression.

Because the present invention is the important evidence that the band limits fallen into by power peak frequency is detected as acoustic feedback, the division of band limits here take Whole frequency band as benchmark, therefore it is without the need to bandpass filter detection signal, also would not occur because the centre frequency of bandpass filter and bandwidth arrange the undetected problem that size causes.The present invention is by the number of crest frequency repetition duration with the decision value satisfied condition in addition, these two values are all whether carry out gauge signal from the angle of the frequency speciality of acoustic feedback be acoustic feedback signal, without the need to going to judge signal period property from the angle of signal waveform, and the computation process of these two values is simply not easily made mistakes, thus avoid the problem that waveform judges the flase drop caused.

Further, the embodiment of the present invention additionally provides the hardware formation of acoustic feedback pick-up unit.At least one processor (such as CPU) can be comprised, at least one network interface or other communication interfaces, storer, and at least one communication bus, for realizing the connection communication between these devices.Processor is used for the executable module stored in execute store, such as computer program.Storer may comprise high-speed random access memory (RAM:Random Access Memory), still may comprise non-labile storer (non-volatile memory), such as at least one magnetic disk memory.Realize the communication connection between this system gateway and at least one other network element by least one network interface (can be wired or wireless), can internet be used, wide area network, local network, Metropolitan Area Network (MAN) etc.

See Fig. 8, in some embodiments, programmed instruction is stored in storer, programmed instruction can be executed by processor, wherein, programmed instruction can comprise time-frequency conversion unit 401, computing unit 402, decision value determining unit 403, decision value determining unit 404 to be counted, acoustic feedback determining unit 405, or programmed instruction also can comprise and suppress unit 406, recognition unit 407, small-signal attenuation units 408, the specific implementation of each unit see Fig. 5,6 or 7 corresponding units disclosed, can repeat no more here.

It should be noted that, acoustic feedback pick-up unit of the present invention can be applied in professional cordless communication network, due to the microphone of intercom and the structure of loudspeaker and its actual application environment, causes intercom easily to occur acoustic feedback phenomenon, therefore, device of the present invention can be applied to intercom.Certainly, the acoustic equipment being connected with microphone and loudspeaker also can be applied to, as in the equipment such as Public Address system, osophone.

It should be noted that, in this article, the such as relational terms of first and second grades and so on is only used for an entity or operation to separate with another entity or operational zone, and not necessarily requires or imply the relation that there is any this reality between these entities or operation or sequentially.And, term " comprises ", " comprising " or its any other variant are intended to contain comprising of nonexcludability, thus make to comprise the process of a series of key element, method, article or equipment and not only comprise those key elements, but also comprise other key elements clearly do not listed, or also comprise by the intrinsic key element of this process, method, article or equipment.When not more restrictions, the key element limited by statement " comprising ... ", and be not precluded within process, method, article or the equipment comprising described key element and also there is other identical element.

The foregoing is only preferred embodiment of the present invention, be not intended to limit protection scope of the present invention.The present invention does not limit the number of time slots and ratio setting relation that can support, and all any amendments done within the spirit and principles in the present invention, equivalent replacement, improvement etc., be all included in protection scope of the present invention.

Claims (10)

1. an acoustic feedback detection method, is characterized in that, comprising:

Time-frequency conversion is carried out to the time-domain signal received and obtains corresponding frequency-region signal;

According to described frequency-region signal determination power peak, calculate the adjacent power of multiple point in described power peak left and right and the average power content of value and described frequency signal;

According to described power and value and described average power content determination decision value;

Determine corresponding preset first threshold according to the band limits that described power peak frequency falls into, if described decision value is greater than described preset first threshold, then determine that described decision value is the decision value to be counted that described band limits is corresponding;

The number of statistics decision value described to be counted that described band limits is corresponding in preset time, and the power peak fallen in preset time in described band limits repeats duration; If described number is greater than preset Second Threshold or described repetition duration when being greater than the 3rd preset threshold value, then determine acoustic feedback is occurring.

2. method according to claim 1, is characterized in that, described determine acoustic feedback to occur after, described method also comprises:

Be decreased to the mode of desired value according to gain coefficient gradually, attenuation processing carried out to described time-domain signal and obtains the signal after acoustic feedback suppression.

3. method according to claim 1, is characterized in that, described method also comprises:

Identify whether the amplitude of described time-domain signal is less than the 4th preset threshold value;

Signal amplitude being less than to the 4th preset threshold value obtains the time-domain signal after small-signal decay as attenuation processing;

The then described mode being decreased to desired value according to gain coefficient gradually, attenuation processing is carried out to described time-domain signal and obtains the signal after acoustic feedback suppression, be specially:

Be decreased to the mode of desired value according to gain coefficient gradually, attenuation processing carried out to the time-domain signal after described small-signal decay and obtains the signal after acoustic feedback suppression.

4. according to the method in any one of claims 1 to 3, it is characterized in that, if described number be less than or equal to preset Second Threshold and described repetition duration is less than or equal to the 3rd preset threshold value time, then determine acoustic feedback does not occur,

Then described method also comprises:

Calculate the maximum likelihood ratio of speech frame according to described time-domain signal and noise power-value, if described maximum likelihood ratio is greater than preset voice threshold value, determine that described time-domain signal is voice signal;

Increase to the mode of 0dB according to described gain coefficient gradually, gain process is carried out to described voice signal and obtains the signal after gain process.

5. according to the method in any one of claims 1 to 3, it is characterized in that, according to described frequency-region signal determination power peak, calculate the adjacent power of multiple point in described power peak left and right and the average power content of value and described frequency signal, specifically comprise:

According to formula the adjacent multiple point in rated output peak value left and right and value, wherein, Xmax (0) is power peak; When j is not equal to 0, Xmax (j) is expressed as the adjacent multiple points in power peak left and right; K is more than or equal to 1;

According to formula $\mathrm{Pav}=[\underset{j=0}{\overset{N/2-1}{\mathrm{\Σ}}}\mathrm{Xj}-\underset{j=-k}{\overset{k}{\mathrm{\Σ}}}X\max \left(j\right)]/(\frac{N}{2}-2k-1)$ Calculate average power content, wherein, represent all performance numbers of described frequency signal power spectrum and value.

6. an acoustic feedback pick-up unit, is characterized in that, comprising:

Time-frequency conversion unit, obtains corresponding frequency-region signal for carrying out time-frequency conversion to the time-domain signal received;

Computing unit, for according to described frequency-region signal determination power peak, calculates the adjacent power of multiple point in described power peak left and right and the average power content of value and described frequency signal;

Decision value determining unit, for according to described power and value and described average power content determination decision value;

Decision value determining unit to be counted, band limits for falling into according to described power peak frequency determines corresponding preset first threshold, if described decision value is greater than described preset first threshold, then determine that described decision value is the decision value to be counted that described band limits is corresponding;

Acoustic feedback determining unit, for adding up the number of decision value described to be counted corresponding to described band limits in preset time, and the power peak fallen in preset time in described band limits repeats duration; If described number is greater than preset Second Threshold or described repetition duration when being greater than the 3rd preset threshold value, then determine acoustic feedback is occurring.

7. device according to claim 7, is characterized in that, described device also comprises:

Suppress unit, for being decreased to the mode of desired value gradually according to gain coefficient, attenuation processing being carried out to described time-domain signal and obtains the signal after acoustic feedback suppression.

8. device according to claim 7, is characterized in that, described device also comprises:

Recognition unit, for identifying whether the amplitude of described time-domain signal is less than the 4th preset threshold value;

Small-signal attenuation units, obtains the time-domain signal after small-signal decay for signal amplitude being less than to the 4th preset threshold value as attenuation processing;

Then described suppression unit, specifically for being decreased to the mode of desired value gradually according to gain coefficient, carrying out attenuation processing to the time-domain signal after described small-signal decay and obtains the signal after acoustic feedback suppression.

9. the device according to any one of claim 6 to 8, is characterized in that, described device also comprises:

Voice determining unit, for calculating the maximum likelihood ratio of speech frame according to described time-domain signal and noise power-value, if described maximum likelihood ratio is greater than preset voice threshold value, determines that described time-domain signal is voice signal;

Speech gain processing unit, for increasing to the mode of 0dB gradually according to described gain coefficient, carries out gain process to described voice signal and obtains the signal after gain process.

10. an acoustic feedback pick-up unit, is characterized in that, comprises at least one processor, at least one network interface or other communication interface, storer, and at least one communication bus; Described storer is used for stored program instruction, and described processor is used for performing following steps according to described programmed instruction:

Time-frequency conversion is carried out to the time-domain signal received and obtains corresponding frequency-region signal;

According to described frequency-region signal determination power peak, calculate the adjacent power of multiple point in described power peak left and right and the average power content of value and described frequency signal;

According to described power and value and described average power content determination decision value;

Determine corresponding preset first threshold according to the band limits that described power peak frequency falls into, if described decision value is greater than described preset first threshold, then determine that described decision value is the decision value to be counted that described band limits is corresponding;

The number of statistics decision value described to be counted that described band limits is corresponding in preset time, and the power peak fallen in preset time in described band limits repeats duration; If described number is greater than preset Second Threshold or described repetition duration when being greater than the 3rd preset threshold value, then determine acoustic feedback is occurring.