CN100483510C - Voice level adjusting device and method - Google Patents
Voice level adjusting device and method Download PDFInfo
- Publication number
- CN100483510C CN100483510C CNB2004100884051A CN200410088405A CN100483510C CN 100483510 C CN100483510 C CN 100483510C CN B2004100884051 A CNB2004100884051 A CN B2004100884051A CN 200410088405 A CN200410088405 A CN 200410088405A CN 100483510 C CN100483510 C CN 100483510C
- Authority
- CN
- China
- Prior art keywords
- level
- energy
- module
- gain
- short
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Images
Landscapes
- Control Of Amplification And Gain Control (AREA)
Abstract
The invention discloses a voice level controller which comprises a fixed level adjusting module, a feedback module, an automatic level adjusting module and an inputting signal adjusting module. The invention also discloses a voice level controlling method. When the system configuration has fixed gain, it dose fixed level adjusting to the inputting signal; when the system configuration dose not have fixed gain or the inputting signal can't achieve the intended requirement after the adjusting of the fixed level, it dose automatic level adjusting to the inputting signal and achieve the automatic level adjusting with changeable rate by the changing rule of the voice during human speaking and changing the accumulating ration of the short time energy into the long time energy.
Description
Technical field
The present invention relates to the voice transfer field, be specifically related to a kind of voice level adjusting device and method.
Background technology
At present, voice quality remains one of problem that the user is concerned about the most.In digital communications network, the transmission of speech data is to carry out between end-to-end, in transmission course, because not matching etc. of Line Attenuation, noise, each network equipment may cause speech level to change, speech level is excessive or too smallly all will finally influence the system voice quality, and those who answer's auditory effect is impacted.Therefore, need the network equipment that speech level is controlled, voice signal is adjusted to the receptible state of people.
The reason that causes speech level to change is a lot, and different reasons causes that the situation of change of speech level also is different, is summarized as follows:
When having level and the unmatched situation of impedance between the communication facilities of different manufacturers, voice will cause the fixed attenuation of speech level when these equipment room transmission, if the amplitude of signal is excessive or too small, can be easy to exceed the dynamic range of the network equipment, thereby cause distorted signals or distortion, can not recognize original sound at receiving end; In addition, in digital wireless network, generally all adopt echo eliminator to eliminate acoustic echo, echo eliminator can cause the fixed attenuation of speech level.
Except above two kinds of situations, following reason can cause the on-fixed decay of speech level:
Voice signal because Line Attenuation, can make voice signal be attenuated or overamplification in transmission course, and voice can be subjected to the influence of noise etc. simultaneously, and the sound heard of receiving end can be very uncomfortable like this, feels aphthenxia or too noisy.
In cordless communication network, in order to save limited bandwidth, at base station sub-system (BSS, BaseStation Subsystem) carries out the preceding voice of wireless signal transmission and must pass through encoding compression, with the data compression of 64Kbit/ second is less than the 16Kbit/ speech parameter data of second, these speech parameter data revert to normal voice after decoding on the mobile phone, oppositely from the mobile phone to BSS too.The Nonlinear Processing of encoding and decoding speech may cause the variation of speech level.
Also there is such problem in the practical application: because regional user's custom speech level difference, the user of different regions is bigger to the demand difference of speech level, in identical wireless network, identical speech level, the user of certain areas reflects again and again that but speech level is too low; On the other hand, there is level differences in different terminals.
By above analysis as can be seen, the speech level adjustment need be finished following function: compensate the caused fixed level decay of some device; Adjust the speech level variation that Line Attenuation, noise etc. cause; Adjust the level differences of different terminals; Satisfy the different demands of regional user to system level.
The level method of adjustment that adopts generally has two kinds at present:
1, single fixed level adjustment
This method is used for the compensatory adjustment of level after fixed attenuation.Whether need to carry out the fixed level adjustment and adjust the size that gains, learn by systematic parameter being tested, analyzing usually.The yield value of adjusting is in case by system configuration, under any circumstance all remain unchanged.
2, the single constant automatic electric-level adjustment of adjustment speed
This method does not adopt when having the fixed attenuation of speech level in system.Calculate the gain of speech level according to the difference of the speech level after the transmission of expectation level and via line, adjust speech level with this.Speech level is that energy or short-time energy draw when calculating the long voice signal of certain window long, and this window is long in case definite, all remains unchanged in whole adjustment process, and it is constant promptly to adjust speed.
Fig. 1 has represented voice level adjusting device possible position in network.As can be seen from Figure 1, speech level adjusting module and voice compression module bind together, the speech level variation that is used to adjust voice coding and decodes and cause; Speech level adjusting module and echo eliminator combine, and are used to compensate the speech level fixed attenuation that echo eliminator causes; Dotted line among the figure represents that the speech level adjusting module may occur on these positions, be used to adjust speech level that Line Attenuation, noise etc. cause and change or adjust the level differences of different terminals and satisfy regional user to the different demands of speech level etc.
The shortcoming of prior art is:
1, the requirement to the system voice level is not satisfied in single fixed level adjustment.The speech level of transmitting on the line owing to system is constantly to change, and the gain of fixed level adjustment is all fixed, and like this, when the adjustment gain is negative, causes little sound can not hear clearly easily; Adjust gain and be timing, cause big sound level value to surpass the scope that the people can tolerate easily, the result has reduced voice quality on the contrary.
2, single automatic electric-level adjustment is not suitable for the compensatory adjustment of level after fixed attenuation.When automatic electric-level is adjusted at the speech level of adjusting through fixed attenuation, because former voice level fluctuates up and down, thereby make when adjusting gain for a short time when big, the result causes the distortion of system voice level.
3, adjust the automatic electric-level adjustment that speed remains unchanged and to follow the tracks of the variation of speech level in time.Adjustment speed is slow, can cause speech level when variation is very fast, delays to follow the tracks of the variation of level; Adjustment speed is fast, under the more stable situation of speech level, emergent level fluctuation is had tangible weakening trend, causes adjusted speech level to reduce.Keeping under the constant situation of adjustment speed, the variation that accomplish the adaptive voice level is the comparison difficulty.
Therefore, adopting single level to adjust mode and single adjustment speed in system not only can't practical requirement, also will introduce some negative factors simultaneously.
Summary of the invention
In view of this, the invention provides a kind of voice level adjusting device and method, satisfying different system and the different demands of different user, and further realize adjustment in time that speech level is changed speech level.
For achieving the above object, technical scheme of the present invention is achieved in that
A kind of voice level adjusting device, this device comprises: fixed level adjusting module, feedback module, automatic electric-level adjusting module and input signal adjusting module, wherein,
The fixed level adjusting module is used to preserve the fixed gain of speech level, and fixed gain is outputed to input signal adjusting module and automatic electric-level adjusting module respectively, when speech level does not have fixed attenuation, and this module no-output;
Feedback module is used for the output of input signal adjusting module is monitored, and when the output of input signal adjusting module does not reach pre-provisioning request, starts the automatic electric-level adjusting module;
The automatic electric-level adjusting module when not receiving the output of fixed level adjusting module, perhaps when receiving the startup of feedback module, according to the input speech signal that receives, calculates the gain that current level is adjusted, and gain is outputed to the input signal adjusting module;
The input signal adjusting module is adjusted input signal according to the output of fixed level adjusting module and automatic electric-level adjusting module, and adjusted input signal is outputed to feedback module.
Described automatic electric-level adjusting module further comprises: energy computing module, expectation level load module, adjustment gain estimation module when short-time energy computing module, noise gate determination module, signal condition judge module, speed adjusting module, length, wherein,
The short-time energy computing module according to the input signal that receives, calculates the short-time energy of input signal and the signal level of this short-time energy correspondence, and short-time energy is outputed to energy computing module when long, and described signal level is outputed to the signal condition judge module;
The noise gate determination module is used to preserve noise gate, and noise gate is outputed to the signal condition judge module;
The signal condition judge module, signal level according to described noise gate and the output of described short-time energy computing module judges that current input signal is voice or noise, if signal level is less than noise gate, judge that then current input signal is a noise, and judged result is outputed to the adjustment gain estimation module; Otherwise the judgement current input signal is voice, starts the speed adjusting module;
The speed adjusting module, when receiving the enabled instruction of described signal condition judge module output, begin accumulated time timing to voice, when the accumulated time of voice reaches the time that the voice rapid adjustment finishes, reduce short-time energy accumulation proportion in the energy when long, and the minimizing value that will accumulate proportion outputs to energy computing module when long, when the accumulated time of voice does not reach the time that the voice rapid adjustment finishes, keeps short-time energy accumulation proportion in the energy when long constant;
Energy computing module when long, according to the short-time energy of described input signal and previous energy when long, and the accumulation proportion minimizing value of speed adjusting module output energy when calculating current speech signal long, energy meter is calculated the current speech level during according to this length, and the current speech level is outputed to the adjustment gain estimation module;
Expectation level load module is used for preserving the expectation level, and will expects that level outputs to the adjustment gain estimation module;
Adjust gain estimation module, according to the current speech level of expectation level with described energy computing module output when long, the difference of calculating both obtains adjusting gain, and will adjust to gain and output to the input signal adjusting module; When receiving the signal condition judge module when being output as noise, keep current adjustment gain constant.
Described automatic electric-level adjusting module further comprises short-time energy maximum value calculation module, this module is used to search for the short-time energy maximal value within a certain period of time of short-time energy computing module output, calculate the signal level of this short-time energy maximal value correspondence, and this signal level is outputed to the noise gate determination module as new noise gate.
Described automatic electric-level adjusting module further comprises clamps down on the level load module, and the maximum that is used to preserve noise gate clamps down on level and minimum is clamped down on level, and maximum is clamped down on level and minimum clamp down on level and output to the noise gate determination module;
Described noise gate determination module further comprises, clamps down on going up in limited time of level when noise gate surpasses, and the value of current noise gate is made as the value of clamping down on the level upper limit; When noise gate be lower than clamp down on level following in limited time, the value of current noise gate is made as the value of clamping down on the level lower limit.
Described automatic electric-level adjusting module comprises that further maximum clamps down on the gain load module, is used to preserve maximum and clamps down on gain, and maximum is clamped down on gain output to the adjustment gain estimation module;
Described adjustment gain estimation module further comprises, when the adjustment gain is clamped down on gain greater than maximum, clamps down on gain with maximum and replaces the adjustment gain.
A kind of speech level method of adjustment, this method comprises, when speech level has fixed attenuation, input signal is carried out the fixed level adjustment, whether judgement reaches pre-provisioning request through the level of the adjusted input signal of fixed level, if reach pre-provisioning request, then keeps current fixed level adjustment; If do not reach pre-provisioning request, input signal is carried out the automatic electric-level adjustment; When speech level does not have fixed attenuation, directly input signal is carried out the automatic electric-level adjustment.
Described automatic electric-level set-up procedure comprises,
101, calculate the short-time energy and the corresponding signal level thereof of input signal;
102, whether the signal level in the determining step 101 less than noise gate, if, judge that then current demand signal is a noise, keep current gain constant, then process ends; Otherwise, the judgement current input signal is voice, and begin the timing of voice accumulated time, judge that whether the voice accumulated time is less than the voice rapid adjustment deadline, if, according to short-time energy and previous when long energy calculate the current current speech level of energy correspondence when energy and this length when long; Otherwise, according to short-time energy, previous when long the minimizing value of energy and short-time energy accumulation proportion in the energy when long calculate the current current speech level of energy correspondence when energy and this length when long;
103, according to adjusting the difference of gain for expectation level and speech level, gain is adjusted in calculating.
In the step 102, described according to short-time energy and previous when long energy calculate current when long when energy and this length the current speech level of energy correspondence comprise, described when long energy equal 1-r on the MAD of r and short-time energy and the previous product of energy when long, wherein, r is short-time energy accumulation proportion in the energy when long;
Described according to short-time energy, previous when long the minimizing value of energy and short-time energy accumulation proportion in the energy when long calculate current when long when energy and this length the current speech level of energy correspondence comprise, described when long energy be 1-r+ Δ r and the previous product of energy when long on the MAD of r-Δ r and short-time energy, wherein, Δ r is the minimizing value of short-time energy accumulation proportion in the energy when long.
Further comprise after the described step 101, before the step 102, follow the tracks of the maximal value of short-time energy in the certain hour, calculate the signal level of this short-time energy maximal value correspondence, and replace original noise gate with this signal level.
Further comprise after the described step 101, before the step 102, relatively noise gate and maximum are clamped down on level, noise gate and minimum and are clamped down on level respectively, if noise gate is clamped down on level greater than maximum, then maximum is clamped down on level as current noise gate, if noise gate is clamped down on level less than minimum, then minimum is clamped down on level as current noise gate, otherwise, keep noise gate constant.
Further comprise after the described step 103 judge whether adjust gain clamps down on gain greater than maximum, if the adjustment gain equals maximum and clamps down on gain; Otherwise it is constant to keep adjusting gain.
Compared with prior art, the wireless system voice level adjusting device that fixed level adjustment and automatic electric-level adjustment are combined provided by the present invention, by the monitoring of feedback module to the speech level of system's output, realized the combination that automatic electric-level adjustment and fixed level are adjusted, simultaneously, when fixed level is adjusted no-output, can realize independently automatic electric-level adjustment, when the feedback module no-output, can realize independently fixed level adjustment, apparatus of the present invention have remedied single level adjustment mode can not satisfy the deficiency that the system voice level requires sometimes; Speech level method of adjustment provided by the invention when system configuration has fixed gain, is carried out the fixed level adjustment to input signal; When system does not dispose fixed gain or when the adjusted input signal of fixed level does not reach pre-provisioning request, input signal is carried out the automatic electric-level adjustment, and Changing Pattern according to voice in people's communication process, by changing short-time energy accumulation proportion in the energy when long, realize the automatic electric-level adjustment of variable bit rate, made the speech level adjustment in time follow the tracks of the variation of speech level.
Description of drawings
Fig. 1 is the position of voice level adjusting device in digital wireless network;
Fig. 2 is the synoptic diagram of voice level adjusting device of the present invention;
Fig. 3 is the composition synoptic diagram of automatic electric-level adjusting module in apparatus of the present invention;
Fig. 4 realizes the process flow diagram that speech level is adjusted for the present invention;
Fig. 5 realizes the process flow diagram that the automatic electric-level of variable bit rate is adjusted for the present invention;
Fig. 6 is the automatic gain change curve synoptic diagram that adopts the automatic electric-level method of adjustment of variable bit rate of the present invention in the specific embodiment.
Embodiment
The present invention is further described in more detail below in conjunction with drawings and the specific embodiments.
Fig. 2 is the synoptic diagram of apparatus of the present invention, and as shown in Figure 2, this device mainly comprises with lower module:
Fixed level adjusting module 201: be used to preserve the fixed gain of speech level, and fixed gain is outputed to automatic electric-level adjusting module 203 and input signal adjusting module 204 respectively, when system does not have fixed attenuation, this module no-output.
Feedback module 202: this module is monitored the output of input signal adjusting module 204, does not reach pre-provisioning request if find the output of input signal adjusting module 204, just enabled instruction is outputed to automatic electric-level adjusting module 203.
Herein " pre-provisioning request " refers to a given level range, just reached pre-provisioning request through the adjusted signal level of fixed level within this scope, just do not reach pre-provisioning request outside this scope.
Automatic electric-level adjusting module 203: according to the input speech signal that receives, calculate the automatic gain that automatic electric-level is adjusted, automatic gain is outputed to input signal adjusting module 204.When not receiving the output of fixed level adjusting module 201, this module starts; Perhaps when receiving the enabled instruction of feedback module 202, this module starts.
Input signal adjusting module 204: be used for receiving inputted signal, and input signal is adjusted, and adjusted input signal is outputed to feedback module 202 according to the fixed gain of fixed level adjusting module 201 output and the automatic gain of automatic electric-level adjusting module 203 outputs.Its output is the result can be expressed as: the output of the input signal+fixed level adjusting module 201 before adjusted input signal=adjustment, or: the output of the input signal+automatic electric-level adjusting module 203 before adjusted input signal=adjustment, or: the output of the output+automatic electric-level adjusting module 203 of the input signal+fixed level adjusting module 201 before adjusted input signal=adjustment.
Fig. 3 is the composition synoptic diagram of automatic electric-level adjusting module in apparatus of the present invention, and as shown in Figure 3, the ingredient of this module comprises:
Short-time energy computing module 301: be used for receiving inputted signal, and the short-time energy of calculating input signal and the signal level of short-time energy correspondence, energy computing module 307 when short-time energy is outputed to short-time energy maximum value calculation module 302 and length respectively, the signal level that it is corresponding outputs to signal condition judge module 305.
Short-time energy maximum value calculation module 302: be used for the maximal value of tracking signal short-time energy within a certain period of time, calculate the signal level of this short-time energy maximal value correspondence, and this signal level is outputed to noise gate determination module 304.
Clamp down on level load module 303: be used to preserve the upper and lower bound of clamping down on level of noise gate, and they are outputed to noise gate determination module 304.
Noise gate determination module 304: be used to preserve initial noise gate, and upgrade current noise gate according to the signal level of short-time energy maximal value correspondence, and clamp down on going up in limited time of level when noise gate surpasses, the value of current noise gate is made as the value of clamping down on the level upper limit; When noise gate be lower than clamp down on level following in limited time, the value of current noise gate is made as the value of clamping down on the level lower limit; Otherwise, keep current noise gate constant, and noise gate outputed to signal condition judge module 305.
Signal condition judge module 305: signal level and noise gate according to 301 outputs of short-time energy computing module judge that input signal is voice or noise.If the level of input signal is greater than noise gate, then current demand signal is a noise, and judged result is outputed to adjustment gain estimation module 309; Otherwise current demand signal is voice, and enabled instruction is outputed to speed adjusting module 306.
Speed adjusting module 306: when receiving the enabled instruction of signal condition judge module, begin the timing of voice accumulated time, when the accumulated time of voice reaches the time that the voice rapid adjustment finishes, just reduce short-time energy accumulation proportion r in the energy when long, and the value Δ r that proportion reduces is outputed to energy computing module 307 when long.
Maximum is clamped down on gain and expectation level load module 308: be used to preserve maximum and clamp down on gain and expect level, maximum is clamped down on to gain and is used to clamp down on the maximal value of adjusting gain, maximum is clamped down on gain and expect that level outputs to adjustment gain estimation module 309.
Adjust gain estimation module 309: according to the expectation level and when long the difference of the current speech level of energy computing module 307 outputs draw and adjust gain, and will adjust to gain and output to input signal adjusting module 310.
Fig. 4 is that the present invention carries out the process flow diagram that speech level is adjusted, and as shown in Figure 4, the step that the present invention carries out the speech level adjustment is as follows:
Whether step 403, judgement reach pre-provisioning request through the adjusted input signal of fixed level, if, execution in step 404; Otherwise, execution in step 405.
Input signal is carried out the automatic electric-level adjustment, can adopt the constant automatic electric-level method of adjustment of adjustment speed of the prior art, also can adopt the automatic electric-level method of adjustment of variable bit rate as shown in Figure 5.
Fig. 5 is the process flow diagram that the present invention carries out the automatic electric-level adjustment of variable bit rate to certain voice signal, and the rapid adjustment deadline of supposing this voice signal is T.As shown in Figure 5, to carry out the step that automatic electric-level adjusts as follows in the present invention:
The short-time energy Es of step 501, signal calculated and corresponding signal level L thereof
Es
The maximal value EsM of step 502, tracking signal short-time energy Es in the certain hour scope, the signal level of establishing the EsM correspondence is L
EsM
Step 503: the noise gate L of update signal
NT=L
EsM
Step 504: compare L respectively
NTClamp down on level L with maximum
NTMax, L
NTClamp down on level L with minimum
NTMin, if L
NTL
NTMax, L then
NT=L
NTMax, and execution in step 506; If L
NT＜L
NTMin, L then
NT=L
NTMin, and execution in step 506; Otherwise, direct execution in step 505.
Step 505: keep L
NTConstant.
Step 506: compare current demand signal level L
EsWith L
NT, if L
Es＜L
NT, execution in step 507; Otherwise, execution in step 508.
Step 507: the judgement current demand signal is a noise, and keeps gain constant, and this flow process finishes.
Step 508: the judgement current demand signal is voice.
Step 509: pick up counting, and whether judge the voice accumulated time less than voice rapid adjustment deadline T, if, execution in step 510; Otherwise, execution in step 511.
Step 510: ENERGY E l (k)=r*Es+ (1-r) * El (k-1) and corresponding speech level L thereof when calculating length
El, energy when wherein El (k) is present segment long, r is short-time energy accumulation proportion in the energy when current segment length, El (k-1) is previous energy when long.Go to step 512 then.
Step 511: ENERGY E l (k)=(r-Δ r) * Es+ (1-r+ Δ r) * El (k-1) and corresponding speech level L thereof when calculating length
ElΔ r is the minimizing value of short-time energy accumulation proportion in the energy when current segment length.Along with the increase of voice time, increase the value of Δ r gradually, reach adjustment speed under the voice steady state (SS) up to the amplitude of Δ r.
Step 512: calculate the value G=L that adjusts gain
e-L
EIWherein, L
eBe the expectation level.
Step 513: relatively G and maximum are clamped down on gain G
Max, if G〉and G
Max, execution in step 514; Otherwise, execution in step 515.
Step 514:G=G
Max
Step 515: keep current G constant.
Fig. 6 is the automatic gain change curve synoptic diagram that adopts the automatic electric-level method of adjustment of variable bit rate of the present invention in the specific embodiment, and the horizontal ordinate of Fig. 6 is the adjustment time, and ordinate is for adjusting gain.As can be seen from Figure 6, automatic gain rises very fast when beginning, and is gradually steady behind the certain hour.This be because: when voice begin, the level of voice signal is less, so in the certain hour after voice begin, short-time energy accumulation proportion in the energy when long is located on the higher value, reach speech level is carried out quick-adjustsing purpose, make voice reach the expectation level within a short period of time; After this speech level is gradually steady, just reduces short-time energy accumulation proportion in the energy when long, with the speed of the speech level adjustment of slowing down.The perception evaluating voice quality test (PESQ of present embodiment, Perceptual Evaluation of SpeechQuality) result is 4.497, the PESQ of former voice signal is 4.5, as can be seen, between voice behind employing the present invention and the former voice very high matching rate is arranged, thereby proved the validity of the inventive method.
The above only is process of the present invention and method embodiment, in order to restriction the present invention, all any modifications of being made within the spirit and principles in the present invention, is not equal to replacement, improvement etc., all should be included within protection scope of the present invention.
Claims (11)
1, a kind of voice level adjusting device is characterized in that, this device comprises: fixed level adjusting module, feedback module, automatic electric-level adjusting module and input signal adjusting module, wherein,
The fixed level adjusting module is used to preserve the fixed gain of speech level, and fixed gain is outputed to input signal adjusting module and automatic electric-level adjusting module respectively, when speech level does not have fixed attenuation, and this module no-output;
Feedback module is used for the output of input signal adjusting module is monitored, and when the output of input signal adjusting module does not reach pre-provisioning request, starts the automatic electric-level adjusting module;
The automatic electric-level adjusting module when not receiving the output of fixed level adjusting module, perhaps when receiving the startup of feedback module, according to the input speech signal that receives, calculates the gain that current level is adjusted, and gain is outputed to the input signal adjusting module;
The input signal adjusting module is adjusted input signal according to the output of fixed level adjusting module and automatic electric-level adjusting module, and adjusted input signal is outputed to feedback module.
2, device as claimed in claim 1, it is characterized in that, described automatic electric-level adjusting module further comprises: energy computing module, expectation level load module, adjustment gain estimation module when short-time energy computing module, noise gate determination module, signal condition judge module, speed adjusting module, length, wherein
The short-time energy computing module according to the input signal that receives, calculates the short-time energy of input signal and the signal level of this short-time energy correspondence, and short-time energy is outputed to energy computing module when long, and described signal level is outputed to the signal condition judge module;
The noise gate determination module is used to preserve noise gate, and noise gate is outputed to the signal condition judge module;
The signal condition judge module, signal level according to described noise gate and the output of described short-time energy computing module judges that current input signal is voice or noise, if signal level is less than noise gate, judge that then current input signal is a noise, and judged result is outputed to the adjustment gain estimation module; Otherwise the judgement current input signal is voice, starts the speed adjusting module;
The speed adjusting module, when receiving the enabled instruction of described signal condition judge module output, begin accumulated time timing to voice, when the accumulated time of voice reaches the time that the voice rapid adjustment finishes, reduce short-time energy accumulation proportion in the energy when long, and the minimizing value that will accumulate proportion outputs to energy computing module when long, when the accumulated time of voice does not reach the time that the voice rapid adjustment finishes, keeps short-time energy accumulation proportion in the energy when long constant;
Energy computing module when long, according to the short-time energy of described input signal and previous energy when long, and the accumulation proportion minimizing value of speed adjusting module output energy when calculating current speech signal long, energy meter is calculated the current speech level during according to this length, and the current speech level is outputed to the adjustment gain estimation module;
Expectation level load module is used for preserving the expectation level, and will expects that level outputs to the adjustment gain estimation module;
Adjust gain estimation module, according to the current speech level of expectation level with described energy computing module output when long, the difference of calculating both obtains adjusting gain, and will adjust to gain and output to the input signal adjusting module; When receiving the signal condition judge module when being output as noise, keep current adjustment gain constant.
3, device as claimed in claim 2, it is characterized in that, described automatic electric-level adjusting module further comprises short-time energy maximum value calculation module, this module is used to search for the short-time energy maximal value within a certain period of time of short-time energy computing module output, calculate the signal level of this short-time energy maximal value correspondence, and this signal level is outputed to the noise gate determination module as new noise gate.
4, device as claimed in claim 2, it is characterized in that, described automatic electric-level adjusting module further comprises clamps down on the level load module, the maximum that is used to preserve noise gate clamps down on level and minimum is clamped down on level, and maximum is clamped down on level and minimum clamp down on level and output to the noise gate determination module;
Described noise gate determination module further comprises, clamps down on going up in limited time of level when noise gate surpasses, and the value of current noise gate is made as the value of clamping down on the level upper limit; When noise gate be lower than clamp down on level following in limited time, the value of current noise gate is made as the value of clamping down on the level lower limit.
5, device as claimed in claim 2 is characterized in that, described automatic electric-level adjusting module comprises that further maximum clamps down on the gain load module, is used to preserve maximum and clamps down on gain, and maximum is clamped down on gain output to the adjustment gain estimation module;
Described adjustment gain estimation module further comprises, when the adjustment gain is clamped down on gain greater than maximum, clamps down on gain with maximum and replaces the adjustment gain.
6, a kind of speech level method of adjustment, it is characterized in that, this method comprises, when speech level has fixed attenuation, input signal is carried out the fixed level adjustment, whether judgement reaches pre-provisioning request through the level of the adjusted input signal of fixed level, if reach pre-provisioning request, then keeps current fixed level adjustment; If do not reach pre-provisioning request, input signal is carried out the automatic electric-level adjustment; When speech level does not have fixed attenuation, directly input signal is carried out the automatic electric-level adjustment.
7, method as claimed in claim 6 is characterized in that, described automatic electric-level set-up procedure comprises,
101, calculate the short-time energy and the corresponding signal level thereof of input signal;
102, whether the signal level in the determining step 101 less than noise gate, if, judge that then current demand signal is a noise, keep current gain constant, then process ends; Otherwise, the judgement current input signal is voice, and begin the timing of voice accumulated time, judge that whether the voice accumulated time is less than the voice rapid adjustment deadline, if, according to short-time energy and previous when long energy calculate the current current speech level of energy correspondence when energy and this length when long; Otherwise, according to short-time energy, previous when long the minimizing value of energy and short-time energy accumulation proportion in the energy when long calculate the current current speech level of energy correspondence when energy and this length when long;
103, according to adjusting the difference of gain for expectation level and speech level, gain is adjusted in calculating.
8, method as claimed in claim 7, it is characterized in that, in the step 102, described according to short-time energy and previous when long energy calculate current when long when energy and this length the current speech level of energy correspondence comprise, described when long energy equal 1-r on the MAD of r and short-time energy and the previous product of energy when long, wherein, r is short-time energy accumulation proportion in the energy when long;
Described according to short-time energy, previous when long the minimizing value of energy and short-time energy accumulation proportion in the energy when long calculate current when long when energy and this length the current speech level of energy correspondence comprise, described when long energy be 1-r+ Δ r and the previous product of energy when long on the MAD of r-Δ r and short-time energy, wherein, Δ r is the minimizing value of short-time energy accumulation proportion in the energy when long.
9, method as claimed in claim 7, it is characterized in that, further comprise after the described step 101, before the step 102, follow the tracks of the maximal value of short-time energy in the certain hour, calculate the signal level of this short-time energy maximal value correspondence, and replace original noise gate with this signal level.
10, method as claimed in claim 7, it is characterized in that, further comprise after the described step 101, before the step 102, relatively noise gate and maximum are clamped down on level, noise gate and minimum and are clamped down on level respectively, if noise gate is clamped down on level greater than maximum, then maximum are clamped down on level as current noise gate, if noise gate is clamped down on level less than minimum, then minimum is clamped down on level as current noise gate, otherwise, keep noise gate constant.
11, method as claimed in claim 7 is characterized in that, further comprises after the described step 103, judges whether adjust gain clamps down on gain greater than maximum, if the adjustment gain equals maximum and clamps down on gain; Otherwise it is constant to keep adjusting gain.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2004100884051A CN100483510C (en) | 2004-10-29 | 2004-10-29 | Voice level adjusting device and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2004100884051A CN100483510C (en) | 2004-10-29 | 2004-10-29 | Voice level adjusting device and method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1766991A CN1766991A (en) | 2006-05-03 |
| CN100483510C true CN100483510C (en) | 2009-04-29 |
Family
ID=36742836
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2004100884051A Expired - Fee Related CN100483510C (en) | 2004-10-29 | 2004-10-29 | Voice level adjusting device and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100483510C (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101132452A (en) * | 2007-07-20 | 2008-02-27 | 华为技术有限公司 | Method and system for adjusting port parameters of voice local side |
| CN101740037B (en) * | 2008-11-21 | 2011-12-28 | 联芯科技有限公司 | Method for controlling automatic gain and dynamic range of voice based on sliding gain |
| CN104575498B (en) * | 2015-01-30 | 2018-08-17 | 深圳市云之讯网络技术有限公司 | Efficient voice recognition methods and system |
| CN110689901B (en) * | 2019-09-09 | 2022-06-28 | 苏州臻迪智能科技有限公司 | Voice noise reduction method and device, electronic equipment and readable storage medium |
| CN112216302B (en) * | 2020-09-09 | 2024-10-18 | 深圳市欢太科技有限公司 | Audio signal processing method, device, electronic equipment and readable storage medium |
-
2004
- 2004-10-29 CN CNB2004100884051A patent/CN100483510C/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| CN1766991A (en) | 2006-05-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US7133529B2 (en) | Howling detecting and suppressing apparatus, method and computer program product | |
| US20030043940A1 (en) | Digital automatic gain control with feedback induced noise suppression | |
| CN101067927B (en) | Sound volume adjusting method and device | |
| EP2385624A1 (en) | Gain control method, gain control equipment in multiple sound channels system, and voice processing system | |
| KR101121212B1 (en) | Method of transmitting data in a communication system | |
| US8363820B1 (en) | Headset with whisper mode feature | |
| KR20120051595A (en) | Intelligibility control using ambient noise detection | |
| CN101478589A (en) | Communication terminal and audio gain parameter regulation method | |
| JPH11340911A (en) | Method and device for controlling power to be radiated by radio terminal in telecommunication system | |
| US20020012324A1 (en) | Echo cancellation with dynamic latency adjustment | |
| CN100483510C (en) | Voice level adjusting device and method | |
| CN101415045A (en) | Method and apparatus for implementing intelligent automatic level control in communication network | |
| US20060014570A1 (en) | Mobile communication terminal | |
| AU680295B2 (en) | Transcoding and transdecoding unit, and method for adjusting the output thereof | |
| US5630014A (en) | Gain controller with automatic adjustment using integration energy values | |
| KR20050063354A (en) | Apparatus and method for frequency controlling considered human's auditory characteristic | |
| US6370366B2 (en) | Telephony device comprising a base station and at least a subscriber unit, subscriber unit for such a telephony device and method used in such a telephony device | |
| CA2026663C (en) | Idle channel noise reduction in a line card having a dsp equalizer | |
| JP3731228B2 (en) | Audio signal transmitting / receiving apparatus and reception volume control method | |
| US7085390B2 (en) | Method for controlling multi-mode audio gain balance | |
| JP3024550B2 (en) | Audio level compensation circuit for mobile terminal equipment | |
| KR100322018B1 (en) | Apparatus for compensating voice band noise | |
| JPH0530161A (en) | Telephone set with automatic adjsuting function for call tone volume | |
| JPH03289743A (en) | Telephone set | |
| AU3785100A (en) | Process and circuit arrangement for echo compensation in a telecommuncations system with non-linear transmission links |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20090429 Termination date: 20201029 |