CN102117618A - Method, device and system for eliminating music noise - Google Patents
Method, device and system for eliminating music noise Download PDFInfo
- Publication number
- CN102117618A CN102117618A CN2009102152120A CN200910215212A CN102117618A CN 102117618 A CN102117618 A CN 102117618A CN 2009102152120 A CN2009102152120 A CN 2009102152120A CN 200910215212 A CN200910215212 A CN 200910215212A CN 102117618 A CN102117618 A CN 102117618A
- Authority
- CN
- China
- Prior art keywords
- noise
- speech signal
- noisy speech
- subband
- signal
- 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.)
- Granted
Links
Images
Landscapes
- Noise Elimination (AREA)
Abstract
The invention discloses a method for eliminating music noise, comprising the following steps of: calculating the signal to noise ratio of a noise-containing speech signal at a low frequency band; modifying the amplitude spectrum estimation parameter of the noise-containing speech signal when the signal to noise ratio meets a preset threshold; and carrying out noise suppression on the noise-containing speech signal by utilizing the modified amplitude spectrum estimation parameter. The embodiment of the invention further provides a device and system for eliminating music noise. Before the noise suppression is carried out, the noise-containing speech signal is detected, the noise-containing speech signal capable of generating music noise is detected, and the amplitude spectrum estimation parameter of the noise-containing speech signal is modified, thus the precision of the detection of the music noise is improved and the music noise is effectively eliminated.
Description
Technical field
The embodiment of the invention relates to the voice communication technical field, relates in particular to a kind of method, Apparatus and system of eliminating the music noise.
Background technology
There are unusual fluctuations at random in the amplitude of nonstationary noise, can utilize the noise suppression algorithm of estimating based on amplitude spectrum to carry out squelch.In the process of noise being carried out the amplitude spectrum estimation, if the noise component of certain frequency is bigger, then can cause estimated result inaccurate, a part of noise is remained, on frequency spectrum, present spike at random, then form the residual noise of rhythmical fluctuating, similar music acoustically, so be called " music noise ".
The music noise that existing a kind of noise reduction techniques based on music noise post processing and filtering, this technology utilize the post-processing technology of squelch to eliminate to produce in the nonstationary noise process of inhibition.The implementation of this technology is to carry out the detection of music noise after to the nonstationary noise noise reduction, then the critical band of detected generation music noise is carried out aftertreatment, to eliminate the music noise.
The inventor finds that there are the following problems at least in the prior art in realizing process of the present invention:
Can lose original statistical property through the nonstationary noise after the noise reduction process, make to cause the remaining music noise characteristic and the feature similarity of voice signal music walkaway precision low, and then can't eliminate the music noise well.
Summary of the invention
Embodiments of the invention provide a kind of method, Apparatus and system of eliminating the music noise, thereby effectively eliminate the music noise.
The objective of the invention is to be achieved through the following technical solutions:
A kind of method of eliminating the music noise comprises:
Calculate the signal to noise ratio (S/N ratio) of noisy speech signal in low-frequency band;
When described signal to noise ratio (S/N ratio) satisfies setting threshold, the amplitude spectrum estimated parameter of described noisy speech signal is revised;
Utilize revised amplitude spectrum estimated parameter described noisy speech signal to be carried out the squelch of estimating based on amplitude spectrum.
A kind of device of eliminating the music noise comprises:
The snr computation module is used to calculate the signal to noise ratio (S/N ratio) of noisy speech signal in low-frequency band;
The parameter correcting module is used for when signal to noise ratio (S/N ratio) that described computing module calculates satisfies setting threshold the amplitude spectrum estimated parameter of described noisy speech signal being revised;
The parameter sending module is used for the amplitude spectrum estimated parameter after the described parameter correcting module processing is sent to the noise suppressor of estimating based on amplitude spectrum.
A kind of system of eliminating the music noise comprises music noise elimination apparatus and noise suppressor:
Described music noise elimination apparatus be used to calculate noisy speech signal in the signal to noise ratio (S/N ratio) of low-frequency band when described signal to noise ratio (S/N ratio) satisfies setting threshold, amplitude spectrum estimated parameter to described noisy speech signal is revised, and revised amplitude spectrum estimated parameter is sent to described noise suppressor;
The revised amplitude spectrum estimated parameter that described noise suppressor is used to utilize described noise suppressor to send carries out the squelch estimated based on amplitude spectrum to described noisy speech signal.
The technical scheme that is provided by the embodiment of the invention described above as can be seen, in the embodiment of the invention, because before carrying out squelch, noisy speech signal is detected, detect the noisy speech signal that may produce the music noise and it is carried out the correction of amplitude spectrum estimated parameter, thereby improved the precision of music walkaway, and then effectively eliminated the music noise.
Description of drawings
In order to be illustrated more clearly in the embodiment of the invention or technical scheme of the prior art, to do one to the accompanying drawing of required use in embodiment or the description of the Prior Art below introduces simply, apparently, accompanying drawing in describing below only is some embodiments of the present invention, for those of ordinary skills, under the prerequisite of not paying creative work, can also obtain other accompanying drawing according to these accompanying drawings.
The processing procedure synoptic diagram that Fig. 1 provides for the embodiment of the invention;
The processing procedure synoptic diagram that Fig. 2 provides for the specific embodiment of the invention;
The apparatus structure synoptic diagram that Fig. 3 provides for the embodiment of the invention;
The system architecture synoptic diagram that Fig. 4 provides for the embodiment of the invention.
Embodiment
Below in conjunction with the accompanying drawing in the embodiment of the invention, the technical scheme in the embodiment of the invention is clearly and completely described, obviously, described embodiment only is the present invention's part embodiment, rather than whole embodiment.Based on the embodiment among the present invention, those of ordinary skills belong to the scope of protection of the invention not making the every other embodiment that is obtained under the creative work prerequisite.
The inventor is through a large amount of analysis of experimentss, and find that the nonstationary noise that produces the music noise possesses following feature: on frequency domain, the nonstationary noise that produces the music noise is distributed in low-frequency band (the upper limiting frequency scope of low-frequency band is at 800 hertz～1100 hertz); On time domain, nonstationary noise can produce the music noise because amplitude spectrum estimates inaccurate when unusual fluctuations occurring.The sub-band division of the voice signal of 8000 hertz of samplings is an example in the following table 1:
| 125-250 | 1250-1437.5 |
| 250-375 | 1437.5-1687.5 |
| 375-500 | 1687.5-1937.5 |
| 500-625 | 1937.5-2250 |
| 625-750 | 2250-2625 |
| 750-875 | 2625-3062.5 |
| 875-1062.5 | 3062.5-3500 |
| 1062.5-1250 | 3500-4000 |
Wherein, noisy speech signal is distributed in 0～4000 hertz, and by being divided into 16 sub-frequency bands shown in the table 1, then preceding 7 sub-frequency bands are low-frequency band (being that the low-frequency band scope is at 0～1062.5 hertz) with noisy speech signal.
The embodiment of the invention provides a kind of music noise cancellation method, feature based on above-mentioned nonstationary noise, before carrying out squelch, noisy speech signal to input is carried out feature detection, judging whether the to produce music noise, and the noisy speech signal that may produce the music noise is carried out making squelch after the correction that amplitude spectrum estimates handle.As shown in Figure 1, this method comprises:
S101, calculate the signal to noise ratio (S/N ratio) of noisy speech signal in low-frequency band;
If this signal to noise ratio (S/N ratio) of S102 satisfies setting threshold, think that then above-mentioned noisy speech signal will produce the music noise, the amplitude spectrum estimated parameter of this noisy speech signal is revised;
S103, utilize revised amplitude spectrum estimated parameter this noisy speech signal to be carried out the squelch of estimating based on amplitude spectrum.
In the above-mentioned processing procedure,, think that then above-mentioned noisy speech signal can not produce the music noise, can directly carry out squelch and handle if this signal to noise ratio (S/N ratio) does not satisfy setting threshold.Wherein, setting threshold comprises lower threshold R
L(span is between 0.5～1.5) and upper limit threshold R
H(span is between 3.5～4.5), signal to noise ratio (S/N ratio) are satisfied setting threshold and are meant that this signal to noise ratio (S/N ratio) is greater than lower threshold R
LAnd less than upper limit threshold R
H
The method that the invention described above embodiment provides, because before carrying out squelch, noisy speech signal is detected, detect the noisy speech signal that may produce the music noise and it is carried out the correction of amplitude spectrum estimated parameter, thereby improved the precision of music walkaway, and then effectively eliminated the music noise.
Among the invention described above embodiment, the amplitude spectrum estimated parameter of noisy speech signal is revised.For example and without limitation, in the application of the noise reduction techniques of estimating based on amplitude spectrum, above-mentioned amplitude spectrum estimated parameter specifically can comprise: the noise energy of noisy speech signal and the gain coefficient of noisy speech signal.Accordingly, can adopt the method for smothing filtering that the amplitude spectrum estimated parameter of noisy speech signal is revised, specifically comprise: the noise energy to above-mentioned noisy speech signal is carried out The disposal of gentle filter; Gain coefficient to above-mentioned noisy speech signal carries out The disposal of gentle filter.To eliminate issuable music noise in the noise suppression process.Wherein, revising preceding amplitude spectrum estimated parameter specifically obtains from noise suppressor.
To be described in detail the specific implementation of the embodiment of the invention in actual application below.
The specific embodiment of the invention provides in a kind of squelch application based on the amplitude spectrum estimation, eliminates the method for music noise, and its concrete processing procedure comprises following operation as shown in Figure 2:
S201, to the input noisy speech signal calculate in the signal to noise ratio (S/N ratio) of low-frequency band;
Signal to noise ratio (S/N ratio) is meant the signal energy that comprises in the above-mentioned noisy speech signal and the ratio of noise energy, wherein, signal energy and noise energy are obtained from noise suppressor, for example and without limitation, comprising the sub-band division module, the subband signal energy estimation block, voice VAD judging module, the noise energy estimation module, subband snr computation module, in the application of the subband noise suppressor of subband gain calculation module, signal energy is from the subband signal energy estimation block of subband noise suppressor, and noise energy obtains from the noise energy estimation module of subband noise suppressor.I frame noisy speech signal is as follows in the computing formula of the signal to noise ratio (S/N ratio) r of low-frequency band (i):
Wherein, m is the sub-band serial number of i frame noisy speech signal, and in above-mentioned subband noise suppressor, noisy speech signal is divided into 16 subbands, and the sub-band division method sees Table 1; E
Ch(m i) is signal energy in m subband of i frame noisy speech signal; E
n(m i) is noise energy in m subband of i frame noisy speech signal; N
LBe the higher limit of low frequency sub-band, the optimal N that the inventor obtains by analysis
LValue is 7 (with table 1 are example, and then Dui Ying frequency is 1062.5 hertz), and the m span is more than or equal to 0 and less than N
LInteger.
S202, judge whether the low-frequency band signal to noise ratio (S/N ratio) that calculates among the S201 satisfies setting threshold,, carry out S203,, then carry out S205 if do not satisfy if satisfy;
Setting threshold comprises lower threshold R
LWith upper limit threshold R
H, specifically can judge whether the low-frequency band signal to noise ratio (S/N ratio) of above-mentioned i frame noisy speech signal satisfies setting threshold by following discriminant:
Wherein, M_flag (i) value is 1 o'clock, is illustrated in through after the noise reduction process of preceding 1～7 subband, and i frame noisy speech signal will produce the music noise; M_flag (i) value is 0 o'clock, represents that i frame noisy speech signal can not produce the music noise.Condition (r
L＜r (i)＜r
H) ﹠amp; ﹠amp; (m＜N
L) (the low frequency sub-band higher limit is N to be illustrated in low-frequency band
L) and r (i) greater than lower threshold R
LAnd less than upper limit threshold R
HThrough a large amount of preferred lower limit threshold value R that obtain that test of inventor
LBe 1, preferred upper limit threshold R
HBe 4.
S203, be that the amplitude spectrum estimated parameter of 1 i frame noisy speech signal is revised to M_flag (i), and carry out S204;
Wherein, in the above-mentioned application based on the subband noise suppressor, the amplitude spectrum estimated parameter specifically comprises: the subband noise energy of i frame noisy speech signal and the subband gain coefficient of i frame noisy speech signal.
The amplitude spectrum estimated parameter revised comprise: 1) the noise energy estimation module from the subband noise suppressor is obtained the subband noise energy, and it is carried out The disposal of gentle filter, and for example and without limitation, the computing formula of The disposal of gentle filter is as follows:
E
N(m, i)=(1-α
0) E
n(m, i-1)+α
0E
n(m, i) (and if only if M_flag (i)==1)
E
N(m, i) expression is through the noise energy of m the subband of i frame noisy speech signal of correction; E
n(m, i-1) noise energy of m the subband of i-1 frame noisy speech signal that provide of expression noise energy estimation module; E
n(m, i) noise energy of m the subband of i frame noisy speech signal that provide of expression noise energy estimation module; α
0Be constant, the acquiescence value is 0.8; 0≤m＜N
L, N
LValue is 15.
2) obtain the subband gain coefficient from the subband gain calculation module of subband noise suppressor, and it is carried out The disposal of gentle filter, for example and without limitation, the computing formula of The disposal of gentle filter is as follows:
g
CH(m，i)＝min{g
ch(m，i-1)，g
ch(m，i)}
g
CH(m, i) expression is through the gain coefficient of m the subband of i frame of correction; g
Ch(m, i-1) gain coefficient of m the subband of i-1 frame that provide of expression subband gain calculation module; g
Ch(m, i) gain coefficient of m the subband of i frame that provide of expression subband gain calculation module.
S204, the squelch of utilizing revised amplitude spectrum estimated parameter that i frame noisy speech signal is carried out estimating based on amplitude spectrum are handled, and this operation is carried out by above-mentioned subband noise suppressor;
S205, the squelch that the i frame noisy speech signal of M_flag (i)=0 is carried out estimating based on amplitude spectrum are handled, and this operation is carried out by above-mentioned subband noise suppressor.
In the above-mentioned processing procedure, can the music noise be set to noisy speech signal and produce sign, and this music noise generation sign is provided with according to the value of M_flag (i).Then in the process of carrying out S203～S204 and S205, can also produce the identical speech signal segments (forming) of sign to the music noise and carry out batch processing by some frame noisy speech signal.
The operation of above-mentioned S201～S203 can be finished by the music noise elimination apparatus of independent setting, also can in the noise suppressor of estimating based on amplitude spectrum, set up the music noise cancellation module and finish, can also carry out software upgrading to the system of the existing noise suppressor of estimating based on amplitude spectrum and finish.By the music noise cancellation method that the invention described above embodiment provides, effectively improved the accuracy of detection of music noise, effectively eliminate the music noise.Especially the implementation by noise suppressor is carried out software upgrading, implementation complexity is low, and need not to increase extra hardware cost.
The all or part of step of realization said method embodiment can be finished by the relevant hardware of programmed instruction, and aforesaid program can be stored in the computer read/write memory medium, and this program is carried out the step that comprises said method embodiment when carrying out; And aforesaid storage medium comprises: various media that can be program code stored such as ROM, RAM, magnetic disc or CD.
The embodiment of the invention also provides a kind of device of eliminating the music noise, feature based on above-mentioned nonstationary noise, before carrying out squelch, noisy speech signal to input is carried out feature detection, to judge whether the to produce music noise, and the noisy speech signal that may produce the music noise is carried out making squelch after the correction that amplitude spectrum estimates handles, this apparatus structure as shown in Figure 3, the specific implementation structure comprises:
Snr computation module 301, be used to calculate the signal to noise ratio (S/N ratio) of noisy speech signal in low-frequency band, wherein, signal to noise ratio (S/N ratio) is meant the signal energy that comprises in the above-mentioned noisy speech signal and the ratio of noise energy, signal energy and noise energy are obtained from noise suppressor, noise energy obtains from the noise energy estimation module of subband noise suppressor, described snr computation module 301 comprises the snr computation submodule, is used to adopt following formula to calculate i frame noisy speech signal at the signal to noise ratio (S/N ratio) r of low-frequency band (i):
Parameter correcting module 302, when the signal to noise ratio (S/N ratio) that calculates when aforementioned calculation module 301 satisfies setting threshold, think that then above-mentioned noisy speech signal will produce the music noise, parameter correcting module 302 is used for the amplitude spectrum estimated parameter of this noisy speech signal is revised;
Parameter sending module 303 is used for the amplitude spectrum estimated parameter after described parameter correcting module 302 processing is sent to the noise suppressor of estimating based on amplitude spectrum, so that carry out noise reduction process.
The device that the invention described above embodiment provides can also comprise judge module (figure does not show), be used to judge whether the signal to noise ratio (S/N ratio) that snr computation module 301 calculates satisfies setting threshold, if this signal to noise ratio (S/N ratio) satisfies setting threshold, then trigger parameter correcting module 302 work, if this signal to noise ratio (S/N ratio) does not satisfy setting threshold, think that then above-mentioned noisy speech signal can not produce the music noise, 303 work of trigger parameter sending module send to noise suppressor by parameter sending module 303 with this noisy speech signal and directly carry out the squelch processing.Wherein, setting threshold comprises lower threshold R
L(span is between 0.5～1.5) and upper limit threshold R
H(span is between 3.5～4.5), signal to noise ratio (S/N ratio) are satisfied setting threshold and are meant that this signal to noise ratio (S/N ratio) is greater than lower threshold R
LAnd less than upper limit threshold R
H
The device that the invention described above embodiment provides, because before carrying out squelch, noisy speech signal is detected, detect the noisy speech signal that may produce the music noise and it is carried out the correction of amplitude spectrum estimated parameter, thereby improved the precision of music walkaway, and then effectively eliminated the music noise.
Among the invention described above embodiment, for example and without limitation, in the application of the noise reduction techniques of estimating based on amplitude spectrum, above-mentioned amplitude spectrum estimated parameter specifically can comprise: the noise energy of noisy speech signal and the gain coefficient of noisy speech signal.Accordingly, parameter correcting module 302 can comprise:
The first parameter correcting module 3021, be used for the noise energy of described noisy speech signal is revised, specifically be used for the noise energy of described noisy speech signal is carried out The disposal of gentle filter, wherein, the noise energy estimation module of noise energy from the subband noise suppressor obtained, the described first parameter correcting module 3021 comprises the first parameter correction submodule, is used to utilize following formula that the subband noise energy of i frame noisy speech signal is carried out The disposal of gentle filter: E
N(m, i)=(1-α
0) E
n(m, i-1)+α
0E
n(m, i) (and if only if M_flag (i)==1) E
N(m, i) expression is through the noise energy of m the subband of i frame noisy speech signal of correction; E
n(m, i-1) noise energy of m the subband of i-1 frame noisy speech signal that provide of expression noise energy estimation module; E
n(m, i) noise energy of m the subband of i frame noisy speech signal that provide of expression noise energy estimation module; α
0Be constant, the acquiescence value is 0.8; 0≤m＜N
L, N
LValue is 15;
The second parameter correcting module 3022, be used for the gain coefficient of described noisy speech signal is revised, specifically be used for the gain coefficient of described noisy speech signal is carried out The disposal of gentle filter, wherein, gain coefficient obtains from the subband gain calculation module of subband noise suppressor, the described second parameter correcting module 3022 comprises the second parameter correction submodule, is used to utilize following formula that the subband gain coefficient of i frame noisy speech signal is carried out The disposal of gentle filter: g
CH(m, i)=min{g
Ch(m, i-1), g
Ch(m, i) } g
CH(m, i) expression is through the gain coefficient of m the subband of i frame of correction; g
Ch(m, i-1) gain coefficient of m the subband of i-1 frame that provide of expression subband gain calculation module; g
Ch(m, i) gain coefficient of m the subband of i frame that provide of expression subband gain calculation module.
Eliminate issuable music noise in the noise suppression process by the mode of smothing filtering.Wherein, revising preceding amplitude spectrum estimated parameter specifically obtains from noise suppressor.
The embodiment of the invention also provides a kind of system of eliminating the music noise, feature based on above-mentioned nonstationary noise, before carrying out squelch, noisy speech signal to input is carried out feature detection, to judge whether the to produce music noise, and the noisy speech signal that may produce the music noise is carried out making squelch after the correction that amplitude spectrum estimates handles, the implementation structure of this system as shown in Figure 4, the specific implementation structure comprises music noise elimination apparatus 401 and noise suppressor 402:
Music noise elimination apparatus 401 be used to calculate noisy speech signal in the signal to noise ratio (S/N ratio) of low-frequency band when described signal to noise ratio (S/N ratio) satisfies setting threshold, amplitude spectrum estimated parameter to described noisy speech signal is revised, and revised amplitude spectrum estimated parameter is sent to described noise suppressor;
The revised amplitude spectrum estimated parameter that noise suppressor 402 is used to utilize described noise suppressor to send carries out the squelch estimated based on amplitude spectrum to described noisy speech signal.
Wherein, the device of the specific implementation structure of the music noise elimination apparatus 401 elimination music noise that can provide with reference to the invention described above embodiment.Noise suppressor 402 can adopt the existing noise suppressor of estimating based on amplitude spectrum.
If above-mentioned signal to noise ratio (S/N ratio) does not satisfy setting threshold, think that then above-mentioned noisy speech signal can not produce the music noise, music noise elimination apparatus 401 sends to noise suppressor 402 with this noisy speech signal and directly carries out the squelch processing.Wherein, setting threshold comprises lower threshold R
LWith upper limit threshold R
H, signal to noise ratio (S/N ratio) satisfies setting threshold and is meant that this signal to noise ratio (S/N ratio) is greater than lower threshold R
LAnd less than upper limit threshold R
H
Among the invention described above embodiment, described noise suppressor 402 also is used for sending to described music noise elimination apparatus 401 the amplitude spectrum estimated parameter of unmodified, described amplitude spectrum estimated parameter comprises: the noise energy of above-mentioned noisy speech signal and the gain coefficient of above-mentioned noisy speech signal.
The system that the invention described above embodiment provides, because before carrying out squelch, noisy speech signal is detected, detect the noisy speech signal that may produce the music noise and it is carried out the correction of amplitude spectrum estimated parameter, thereby improved the precision of music walkaway, and then effectively eliminated the music noise.
The above; only for the preferable embodiment of the present invention, but protection scope of the present invention is not limited thereto, and anyly is familiar with those skilled in the art in the technical scope that the present invention discloses; the variation that can expect easily or replacement all should be encompassed within protection scope of the present invention.Therefore, protection scope of the present invention should be as the criterion with the protection domain of claim.
Claims (13)
1. a method of eliminating the music noise is characterized in that, comprising:
Calculate the signal to noise ratio (S/N ratio) of noisy speech signal in low-frequency band;
When described signal to noise ratio (S/N ratio) satisfies setting threshold, the amplitude spectrum estimated parameter of described noisy speech signal is revised;
Utilize revised amplitude spectrum estimated parameter described noisy speech signal to be carried out the squelch of estimating based on amplitude spectrum.
2. method according to claim 1, it is characterized in that, described signal to noise ratio (S/N ratio) is meant the signal energy that comprises in the described noisy speech signal and the ratio of noise energy, and when described noisy speech signal was unit with the frame, i frame noisy speech signal was as follows at the snr computation formula of low-frequency band:
3. method according to claim 2 is characterized in that the nonstationary noise of described noisy speech signal is distributed in low-frequency band, and the upper limiting frequency scope of described low-frequency band is 800～1100 hertz.
4. method according to claim 2 is characterized in that, this method also comprises: whether the low-frequency band signal to noise ratio (S/N ratio) of judging described i frame noisy speech signal by following discriminant satisfies setting threshold:
Wherein, M_flag (i) value is 1 o'clock, is illustrated in through after the noise reduction process of preceding 1～7 subband, and i frame noisy speech signal will produce the music noise; M_flag (i) value is 0 o'clock, represents that i frame noisy speech signal can not produce the music noise; Condition (r
L＜r (i)＜r
H) ﹠amp; ﹠amp; (m＜N
L) be illustrated in low-frequency band and r (i) greater than lower threshold R
LAnd less than upper limit threshold R
H
5. method according to claim 4, it is characterized in that, described signal to noise ratio (S/N ratio) satisfies setting threshold and is meant: described signal to noise ratio (S/N ratio) is greater than lower threshold and less than upper limit threshold, and the span of described lower threshold is 0.5～1.5, and the span of described upper limit threshold is 3.5～4.5.
6. according to any described method of claim 1～5, it is characterized in that described amplitude spectrum estimated parameter to described noisy speech signal is revised and comprised:
Noise energy to described noisy speech signal is carried out The disposal of gentle filter;
Gain coefficient to described noisy speech signal carries out The disposal of gentle filter.
7. method according to claim 6 is characterized in that, the noise energy of described noisy speech signal is carried out The disposal of gentle filter comprise: utilize following formula that the subband noise energy of i frame noisy speech signal is carried out The disposal of gentle filter:
E
N(m, i)=(1-a
0) E
n(m, i-1)+a
0E
n(m, i) (and if only if M_flag (i)==1),
E
N(m, i) expression is through the noise energy of m the subband of i frame noisy speech signal of correction; E
n(m, i-1) noise energy of m the subband of i-1 frame noisy speech signal that provide of expression noise energy estimation module; E
n(m, i) noise energy of m the subband of i frame noisy speech signal that provide of expression noise energy estimation module; α
0Be constant; 0≤m＜N
L
The gain coefficient of described noisy speech signal is carried out The disposal of gentle filter to be comprised: utilize following formula that the subband gain coefficient of i frame noisy speech signal is carried out The disposal of gentle filter:
g
CH(m，i)＝min(g
ch(m，i-1)，g
ch(m，i)}
g
CH(m, i) expression is through the gain coefficient of m the subband of i frame of correction; g
Ch(m, i-1) gain coefficient of m the subband of i-1 frame that provide of expression subband gain calculation module; g
Ch(m, i) gain coefficient of m the subband of i frame that provide of expression subband gain calculation module.
8. a device of eliminating the music noise is characterized in that, comprising:
The snr computation module is used to calculate the signal to noise ratio (S/N ratio) of noisy speech signal in low-frequency band;
The parameter correcting module is used for when signal to noise ratio (S/N ratio) that described computing module calculates satisfies setting threshold the amplitude spectrum estimated parameter of described noisy speech signal being revised;
The parameter sending module is used for the amplitude spectrum estimated parameter after the described parameter correcting module processing is sent to the noise suppressor of estimating based on amplitude spectrum.
9. device according to claim 8 is characterized in that, described snr computation module comprises:
The snr computation submodule is used to adopt following formula to calculate i frame noisy speech signal at the signal to noise ratio (S/N ratio) r of low-frequency band (i):
10. device according to claim 8 is characterized in that, described parameter correcting module comprises:
The first parameter correcting module is used for the noise energy of described noisy speech signal is carried out The disposal of gentle filter;
The second parameter correcting module is used for the gain coefficient of described noisy speech signal is carried out The disposal of gentle filter.
11. device according to claim 10 is characterized in that, the described first parameter correcting module comprises: the first parameter correction submodule is used to utilize following formula that the subband noise energy of i frame noisy speech signal is carried out The disposal of gentle filter:
E
N(m, i)=(1-a
0) E
n(m, i-1)+α
0E
n(m, i) (and if only if M_flag (i)==1),
Wherein, E
N(m, i) expression is through the noise energy of m the subband of i frame noisy speech signal of correction; E
n(m, i-1) noise energy of m the subband of i-1 frame noisy speech signal that provide of expression noise energy estimation module; E
n(m, i) noise energy of m the subband of i frame noisy speech signal that provide of expression noise energy estimation module; α
0Be constant; 0≤m＜N
L
The described second parameter correcting module comprises: the second parameter correction submodule is used to utilize following formula that the subband gain coefficient of i frame noisy speech signal is carried out The disposal of gentle filter:
g
CH(m，i)＝min{g
ch(m，i-1)，g
ch(m，i)}，
Wherein, g
CH(m, i) expression is through the gain coefficient of m the subband of i frame of correction; g
Ch(m, i-1) gain coefficient of m the subband of i-1 frame that provide of expression subband gain calculation module; g
Ch(m, i) gain coefficient of m the subband of i frame that provide of expression subband gain calculation module.
12. a system of eliminating the music noise is characterized in that, comprises music noise elimination apparatus and noise suppressor:
Described music noise elimination apparatus be used to calculate noisy speech signal in the signal to noise ratio (S/N ratio) of low-frequency band when described signal to noise ratio (S/N ratio) satisfies setting threshold, amplitude spectrum estimated parameter to described noisy speech signal is revised, and revised amplitude spectrum estimated parameter is sent to described noise suppressor;
The revised amplitude spectrum estimated parameter that described noise suppressor is used to utilize described noise suppressor to send carries out the squelch estimated based on amplitude spectrum to described noisy speech signal.
13. system according to claim 12, it is characterized in that, described noise suppressor also is used for sending to described music noise elimination apparatus the amplitude spectrum estimated parameter of unmodified, described amplitude spectrum estimated parameter comprises: the noise energy of described noisy speech signal and the gain coefficient of described noisy speech signal.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN200910215212A CN102117618B (en) | 2009-12-30 | 2009-12-30 | Method, device and system for eliminating music noise |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN200910215212A CN102117618B (en) | 2009-12-30 | 2009-12-30 | Method, device and system for eliminating music noise |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102117618A true CN102117618A (en) | 2011-07-06 |
| CN102117618B CN102117618B (en) | 2012-09-05 |
Family
ID=44216349
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN200910215212A Active CN102117618B (en) | 2009-12-30 | 2009-12-30 | Method, device and system for eliminating music noise |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102117618B (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2013178139A1 (en) * | 2012-08-02 | 2013-12-05 | 中兴通讯股份有限公司 | Signal processing method, apparatus, and electronic device |
| CN103745729A (en) * | 2013-12-16 | 2014-04-23 | 深圳百科信息技术有限公司 | Audio de-noising method and audio de-noising system |
| CN104364845A (en) * | 2012-05-01 | 2015-02-18 | 株式会社理光 | Processing apparatus, processing method, program, computer readable information recording medium and processing system |
| CN108269567A (en) * | 2018-01-23 | 2018-07-10 | 北京百度网讯科技有限公司 | For generating the method, apparatus of far field voice data, computing device and computer readable storage medium |
| CN109671309A (en) * | 2018-12-12 | 2019-04-23 | 广东小天才科技有限公司 | A kind of mistake pronunciation recognition methods and electronic equipment |
| CN110431625A (en) * | 2019-06-21 | 2019-11-08 | 深圳市汇顶科技股份有限公司 | Speech detection method, speech detection device, speech processing chip and electronic equipment |
| CN110458782A (en) * | 2019-08-15 | 2019-11-15 | 深圳市瑞立视多媒体科技有限公司 | Three-dimensional track smoothing processing method, device, equipment and storage medium |
| CN111031186A (en) * | 2019-12-03 | 2020-04-17 | 苏宁云计算有限公司 | Noise processing method, server and client |
| CN111192573A (en) * | 2018-10-29 | 2020-05-22 | 宁波方太厨具有限公司 | Equipment intelligent control method based on voice recognition |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107123419A (en) * | 2017-05-18 | 2017-09-01 | 北京大生在线科技有限公司 | The optimization method of background noise reduction in the identification of Sphinx word speeds |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6122384A (en) * | 1997-09-02 | 2000-09-19 | Qualcomm Inc. | Noise suppression system and method |
| US6088668A (en) * | 1998-06-22 | 2000-07-11 | D.S.P.C. Technologies Ltd. | Noise suppressor having weighted gain smoothing |
| WO2002080148A1 (en) * | 2001-03-28 | 2002-10-10 | Mitsubishi Denki Kabushiki Kaisha | Noise suppressor |
| US7492889B2 (en) * | 2004-04-23 | 2009-02-17 | Acoustic Technologies, Inc. | Noise suppression based on bark band wiener filtering and modified doblinger noise estimate |
| US8204754B2 (en) * | 2006-02-10 | 2012-06-19 | Telefonaktiebolaget L M Ericsson (Publ) | System and method for an improved voice detector |
-
2009
- 2009-12-30 CN CN200910215212A patent/CN102117618B/en active Active
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104364845B (en) * | 2012-05-01 | 2017-03-08 | 株式会社理光 | Processing meanss, processing method, program, computer-readable information recording medium and processing system |
| CN104364845A (en) * | 2012-05-01 | 2015-02-18 | 株式会社理光 | Processing apparatus, processing method, program, computer readable information recording medium and processing system |
| CN103581447B (en) * | 2012-08-02 | 2016-09-07 | 中兴通讯股份有限公司 | A kind of method of signal transacting, device and electronic equipment |
| CN103581447A (en) * | 2012-08-02 | 2014-02-12 | 中兴通讯股份有限公司 | Method and device for signal processing and electronic equipment |
| WO2013178139A1 (en) * | 2012-08-02 | 2013-12-05 | 中兴通讯股份有限公司 | Signal processing method, apparatus, and electronic device |
| CN103745729A (en) * | 2013-12-16 | 2014-04-23 | 深圳百科信息技术有限公司 | Audio de-noising method and audio de-noising system |
| CN103745729B (en) * | 2013-12-16 | 2017-01-04 | 深圳百科信息技术有限公司 | A kind of audio frequency denoising method and system |
| CN108269567A (en) * | 2018-01-23 | 2018-07-10 | 北京百度网讯科技有限公司 | For generating the method, apparatus of far field voice data, computing device and computer readable storage medium |
| CN111192573A (en) * | 2018-10-29 | 2020-05-22 | 宁波方太厨具有限公司 | Equipment intelligent control method based on voice recognition |
| CN111192573B (en) * | 2018-10-29 | 2023-08-18 | 宁波方太厨具有限公司 | Intelligent control method for equipment based on voice recognition |
| CN109671309A (en) * | 2018-12-12 | 2019-04-23 | 广东小天才科技有限公司 | A kind of mistake pronunciation recognition methods and electronic equipment |
| CN110431625A (en) * | 2019-06-21 | 2019-11-08 | 深圳市汇顶科技股份有限公司 | Speech detection method, speech detection device, speech processing chip and electronic equipment |
| CN110458782A (en) * | 2019-08-15 | 2019-11-15 | 深圳市瑞立视多媒体科技有限公司 | Three-dimensional track smoothing processing method, device, equipment and storage medium |
| CN110458782B (en) * | 2019-08-15 | 2022-05-03 | 深圳市瑞立视多媒体科技有限公司 | Three-dimensional track smoothing method, device, equipment and storage medium |
| CN111031186A (en) * | 2019-12-03 | 2020-04-17 | 苏宁云计算有限公司 | Noise processing method, server and client |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102117618B (en) | 2012-09-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102117618B (en) | Method, device and system for eliminating music noise | |
| CN101599269B (en) | Phonetic end point detection method and device therefor | |
| CN101770775B (en) | Signal processing method and device | |
| CN101770779B (en) | Noise spectrum tracking in noisy acoustical signals | |
| WO2018213565A2 (en) | Artificial intelligence-based text-to-speech system and method | |
| US8954320B2 (en) | System and method for noise reduction in processing speech signals by targeting speech and disregarding noise | |
| Ganapathy et al. | Robust feature extraction using modulation filtering of autoregressive models | |
| Xiao et al. | Normalization of the speech modulation spectra for robust speech recognition | |
| CN101968781B (en) | Method of making a window type decision based on MDCT data in audio encoding | |
| CN103325380A (en) | Gain post-processing for signal enhancement | |
| CN104392718A (en) | Robust voice recognition method based on acoustic model array | |
| CN110111811B (en) | Audio signal detection method, device and storage medium | |
| EP3349213B1 (en) | System and method for noise estimation with music detection | |
| Abdullah et al. | Towards more efficient DNN-based speech enhancement using quantized correlation mask | |
| EP3147900B1 (en) | Method and device for processing audio signal | |
| CN103295580A (en) | Method and device for suppressing noise of voice signals | |
| JP2007293059A (en) | Signal processing apparatus and its method | |
| Loweimi et al. | Robust Source-Filter Separation of Speech Signal in the Phase Domain. | |
| CN102598126A (en) | Information processing device, auxiliary device therefor, information processing system, control method therefor, and control program | |
| Joy et al. | Deep Scattering Power Spectrum Features for Robust Speech Recognition. | |
| CN107919136B (en) | Digital voice sampling frequency estimation method based on Gaussian mixture model | |
| US9349383B2 (en) | Audio bandwidth dependent noise suppression | |
| Shao et al. | A versatile speech enhancement system based on perceptual wavelet denoising | |
| Sharma et al. | Speech denoising using different types of filters | |
| Choi | Noise robust front-end for ASR using spectral subtraction, spectral flooring and cumulative distribution mapping |
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 |