CN106531175B - A kind of method that network phone comfort noise generates - Google Patents
A kind of method that network phone comfort noise generates Download PDFInfo
- Publication number
- CN106531175B CN106531175B CN201610996520.1A CN201610996520A CN106531175B CN 106531175 B CN106531175 B CN 106531175B CN 201610996520 A CN201610996520 A CN 201610996520A CN 106531175 B CN106531175 B CN 106531175B
- Authority
- CN
- China
- Prior art keywords
- noise
- voice
- signal
- network phone
- decoding
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 33
- 230000003044 adaptive effect Effects 0.000 claims abstract description 7
- 238000001514 detection method Methods 0.000 claims abstract description 7
- 230000000694 effects Effects 0.000 claims abstract description 3
- 230000003595 spectral effect Effects 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 5
- 230000006835 compression Effects 0.000 claims description 5
- 238000007906 compression Methods 0.000 claims description 5
- 238000006243 chemical reaction Methods 0.000 claims description 2
- 238000001228 spectrum Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 238000004891 communication Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 210000003739 neck Anatomy 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000005236 sound signal Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L19/00—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
- G10L19/012—Comfort noise or silence coding
Landscapes
- Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Signal Processing (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
Abstract
The present invention discloses a kind of method that network phone comfort noise generates, under the feelings for not changing standard agreement, using joined random adaptive codebook and random fixed codebook in white noise, it whether is active speech or nonactive voice by detection load packet signal source, voice signal model is being generated after noise decoding and linear predictive coding calculate, soft noise is generated using linear prediction filter, the effect of the technical program can preferably reflect the ambient noise of actual environment, make acoustically to feel with stability.
Description
Technical field
The present invention relates to audio digital signals processing system in a kind of network telephone more particularly to a kind of network telephones
The method that middle comfort noise generates, the invention belongs to embedded computer system, network communication, media information processing technique necks
Domain.
Background technique
International Telecommunication Union's communication standardization tissue publication audio coder & decoder (codec) technical standard ITU-T G.711
In Appendix II annex 2 of agreement (ITU-T G.711), the production method of two kinds of comfort noises is defined, it is international electricity
Letter alliance ITU-T makes a set of voice compression come, and represents logarithm PCM (logarithmic pulse-code
Modulation) sampling standard；It is mainly used for phone, using pulse code modulation to audio sample, sample rate is that 8k is per second, benefit
With the uncompressed channel transfer speech sound signal of a 64Kbps, 16 data are compressed into 8, G.711 marked by compression ratio 1:2
Standard is the waveform sound codec of mainstream, and G.711 there are mainly two types of compression algorithms under standard, is described below respectively:
As shown in Figure 1, traditional noise generation method schematic diagram with noise gain and frequency spectrum parameter, first method are
The voice signal that payload length is 11 bytes is sent, wherein the 1st byte is the gain parameter of comfort noise, subsequent 10 words
Section be noise frequency spectrum parameter, in receiving end, as long as from load packet in decoding obtain noise gain and frequency spectrum parameter it is linearly pre-
It surveys code coefficient and comfort noise can be obtained using random white noise as driving source.
As shown in Fig. 2, traditional only noise generation method schematic diagram with noise gain, second method are to send load
Length only has the voice signal of 1 byte, only includes the gain parameter of comfort noise in load packet, without mentioning in first method
The frequency spectrum parameter arrived.
In order to reduce the capacity of load packet, so what is actually used at present is all the comfort noise that second method generates,
Since the comfort noise generated under 1 byte mode does not have frequency spectrum parameter, so " soft " noise that practical this method generates is simultaneously
It is not soft.
In addition, in above-mentioned two ways, what is used due to the driving source of comfort noise is all white noise, and use is white
Noise is also bad as comfort noise effect caused by driving source, acoustically also there is discontinuous feeling.
Therefore, in this case, it is proposed that a kind of method that comfort noise generates in new network phone, solves to make an uproar
The stability problem of sound effective value.
Summary of the invention
In order to solve the drawbacks described above of the prior art, the technical program purpose is, under the feelings for not changing standard agreement,
It whether is living by detection load packet signal source using joined random adaptive codebook and random fixed codebook in white noise
Property voice or nonactive voice, generating voice signal model after noise decoding and linear predictive coding calculate, using
Linear Prediction filter is generated than milder noise, can preferably be reflected the ambient noise of actual environment, be made acoustically to feel have
There is stability.
The purpose of the present invention is achieved through the following technical solutions:
A kind of method that network phone comfort noise generates, it is characterised in that: in the voice recognition processing mistake of network phone
Whether Cheng Zhong, detection load packet signal source are active speech or nonactive voice, and the noise of nonactive voice is decoded by noise
Afterwards, so that output noise band is changed spectral characteristic, then enter linear prediction filter；One group of random adaptive code of addition is set
The exciting signal source of this and random fixed codebook, the output noise and exciting signal source that band changes spectral characteristic are by linear pre-
Survey filter；Meanwhile voice output of the voice signal that is obtained after tone decoding of active speech as vocoder, Huo Zhezuo
Comfort noise signal is exported after linear prediction filter for the voice input that linear predictive coding calculates.
The present invention has unique advantages and beneficial effects as follows:
The method that a kind of network phone comfort noise that the technical program proposes generates, before not changing the capacity that load is wrapped
It puts, using the signal for the generation model for more meeting voice signal as driving source, so that the comfort noise of output is more comfortable,
Improve the user experience of voice communication.
Detailed description of the invention
Fig. 1 is traditional noise generation method schematic diagram with noise gain and frequency spectrum parameter；
Fig. 2 is traditional only noise generation method schematic diagram with noise gain；
Fig. 3 is the architecture diagram for the method that a kind of network phone comfort noise of the present invention generates；
Fig. 4 is the calculation formula of the noise gain for the method that a kind of network phone comfort noise of the present invention generates.
Specific embodiment
Technical solution of the present invention is described in detail with reference to the accompanying drawings of the specification:
As shown in figure 3, a kind of method that network phone comfort noise generates, it is characterised in that: in the voice of network phone
In identification processing procedure, whether detection load packet signal source is active speech or nonactive voice, the noise warp of nonactive voice
After crossing noise decoding, output noise band is made to change spectral characteristic, then enters linear prediction filter；Be arranged one group of addition with
The exciting signal source of machine adaptive codebook and random fixed codebook, band change the output noise and exciting signal source of spectral characteristic
By linear prediction filter；Meanwhile voice of the voice signal that is obtained after tone decoding of active speech as vocoder
Output, or the voice input calculated as linear predictive coding, after linear prediction filter, output comfort noise letter
Number.
Further, during the voice recognition processing of network phone, receive load source speech signal, according to load packet
Length detection judges whether load packet signal source is active speech signal, and setting noise as the length of nonactive speech payloads packet is
1 byte is 0 byte, when for 1 byte, decodes and obtains noise gain, when for 0 byte, then the noise gain of present frame
It is constant, then it is replaced with the noise gain of previous frame, and the length of active speech load packet is 160 bytes, for example active speech frame,
It is then transferred to tone decoding, otherwise turns noise decoding.
Further, active speech load packet is transferred to tone decoding, using the compress speech side in G.711 standard
Formula, speech decoding process are one group of obtained voice signal from μ rate or A rate to the conversion of Linear Pulse Code Modulation, a side
Output of the face as vocoder, current frame speech decoding terminates, then turns to do next frame decoding；On the other hand it is compiled as linear prediction
The voice input that code device calculates.
Further, the noise of the nonactive voice carries out noise decoding, is the noise energy solution that will be loaded in packet
Code, obtains noise gain G, using audio coder & decoder (codec) technical standard ITU-T G.711 II protocol mode of Appendix, meter
Calculation mode isWherein E is the noise energy that load decoding obtains.
And then, the voice signal that the active speech obtains after tone decoding is as Linear Predictive Coder meter
The voice of calculation inputs, and determines linear forecast coding coefficient using CELP QCELP Qualcomm.
Then, one group of exciting signal source that random adaptive codebook and random fixed codebook are added in white noise is set,
Using the International Telecommunication Union voice compression algorithm ITU-T calculation method that G.729 B.4.4 annex B agreement saves, according to
Audio coder & decoder (codec) technical standard G.711 in the length of every frame determine, obtain e [80] sequence as output voice signal.
Finally, using the e [80] obtained after calculating as driving source, the noise gain G that noise decodes, by linear
The voice signal that predictive coding is calculated is filtered by linear prediction filter, obtains comfort noise output signal.
In conclusion the method that a kind of network phone comfort noise proposed by the present invention generates, joined in white noise
Random adaptive codebook and random fixed codebook generate voice signal model, so that the comfort noise of output is softer.
Above-mentioned technical proposal is only specific application example of the invention, can be drunk as the case may be in actual application
Feelings select alternate device device, but are not limited in any way to the protection scope of invention.
Claims (6)
1. a kind of method that network phone comfort noise generates, it is characterised in that: in the voice recognition processing process of network phone
In, whether detection load packet signal source is active speech or nonactive voice, the noise of nonactive voice after noise decodes,
So that output noise band is changed spectral characteristic, then enters linear prediction filter；One group of random adaptive codebook of addition is set
With the exciting signal source of random fixed codebook, assisted using International Telecommunication Union voice compression algorithm ITU-TG.729annexB
Discuss the calculation method that B.4.4 saves, according to audio coder & decoder (codec) technical standard G.711 in the length of every frame determine, obtain e
[80] sequence is as output voice signal, and band changes the output noise of spectral characteristic and exciting signal source is filtered by linear prediction
Wave device；Meanwhile voice output of the voice signal that is obtained after tone decoding of active speech as vocoder, or as line
Property predictive coding calculate voice input, after linear prediction filter, export comfort noise signal.
2. the method that a kind of network phone comfort noise according to claim 1 generates, it is characterised in that: network phone
During voice recognition processing, receive load source speech signal, is according to the length detection judgement load packet signal source of load packet
No is active speech signal, sets noise and is 1 byte as the length of nonactive speech payloads packet or is 0 byte, when for 1 byte
When, decoding obtains noise gain, and when for 0 byte, then the noise gain of present frame is constant, then uses the noise gain generation of previous frame
It replaces, and the length of active speech load packet is 160 bytes, for example active speech frame is then transferred to tone decoding, otherwise turns noise solution
Code.
3. the method that a kind of network phone comfort noise according to claim 1 generates, it is characterised in that: the load
Packet signal source is detected as being transferred to tone decoding when active speech, and speech decoding process is encoded from μ rate or A rate to linear impulsive
The conversion of modulation, one group of obtained voice signal, on the one hand as the output of vocoder, current frame speech decoding terminates, then turns
Do next frame decoding；On the other hand the voice input calculated as linear predictive coding.
4. the method that a kind of network phone comfort noise according to claim 1 generates, it is characterised in that: described is non-live
Property voice noise carry out noise decoding, be will load packet in noise energy decode, noise gain G is obtained, using voice coder
Decoder technique standard agreement mode.
5. the method that a kind of network phone comfort noise according to claim 1 generates, it is characterised in that: the activity
The voice input that the voice signal that voice obtains after tone decoding is calculated as Linear Predictive Coder, is swashed using CELP code
Linear predictive coding is encouraged to determine linear forecast coding coefficient.
6. the method that a kind of network phone comfort noise according to claim 1 generates, it is characterised in that: after calculating
Obtained e [80] is used as driving source, the noise gain G that noise decodes, the voice being calculated by linear predictive coding
Signal is filtered by linear prediction filter, obtains comfort noise output signal.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610996520.1A CN106531175B (en) | 2016-11-13 | 2016-11-13 | A kind of method that network phone comfort noise generates |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610996520.1A CN106531175B (en) | 2016-11-13 | 2016-11-13 | A kind of method that network phone comfort noise generates |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106531175A CN106531175A (en) | 2017-03-22 |
| CN106531175B true CN106531175B (en) | 2019-09-03 |
Family
ID=58351327
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610996520.1A Active CN106531175B (en) | 2016-11-13 | 2016-11-13 | A kind of method that network phone comfort noise generates |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106531175B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10373630B2 (en) * | 2017-03-31 | 2019-08-06 | Intel Corporation | Systems and methods for energy efficient and low power distributed automatic speech recognition on wearable devices |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050203733A1 (en) * | 2004-03-15 | 2005-09-15 | Ramkummar Permachanahalli S. | Method of comfort noise generation for speech communication |
| CN101339767A (en) * | 2008-03-21 | 2009-01-07 | 华为技术有限公司 | Background noise excitation signal generating method and apparatus |
| CN101632119A (en) * | 2007-03-05 | 2010-01-20 | 艾利森电话股份有限公司 | Method and arrangement for smoothing of stationary background noise |
| CN102930872A (en) * | 2012-11-05 | 2013-02-13 | 深圳广晟信源技术有限公司 | Method and device for postprocessing pitch enhancement in broadband speech decoding |
| CN103137133A (en) * | 2011-11-29 | 2013-06-05 | 中兴通讯股份有限公司 | In-activated sound signal parameter estimating method, comfortable noise producing method and system |
| CN103680509A (en) * | 2013-12-16 | 2014-03-26 | 重庆邮电大学 | Method for discontinuous transmission of voice signals and generation of background noise |
| CN104978970A (en) * | 2014-04-08 | 2015-10-14 | 华为技术有限公司 | Noise signal processing and generation method, encoder/decoder and encoding/decoding system |
-
2016
- 2016-11-13 CN CN201610996520.1A patent/CN106531175B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050203733A1 (en) * | 2004-03-15 | 2005-09-15 | Ramkummar Permachanahalli S. | Method of comfort noise generation for speech communication |
| CN101632119A (en) * | 2007-03-05 | 2010-01-20 | 艾利森电话股份有限公司 | Method and arrangement for smoothing of stationary background noise |
| CN101339767A (en) * | 2008-03-21 | 2009-01-07 | 华为技术有限公司 | Background noise excitation signal generating method and apparatus |
| CN103137133A (en) * | 2011-11-29 | 2013-06-05 | 中兴通讯股份有限公司 | In-activated sound signal parameter estimating method, comfortable noise producing method and system |
| CN102930872A (en) * | 2012-11-05 | 2013-02-13 | 深圳广晟信源技术有限公司 | Method and device for postprocessing pitch enhancement in broadband speech decoding |
| CN103680509A (en) * | 2013-12-16 | 2014-03-26 | 重庆邮电大学 | Method for discontinuous transmission of voice signals and generation of background noise |
| CN104978970A (en) * | 2014-04-08 | 2015-10-14 | 华为技术有限公司 | Noise signal processing and generation method, encoder/decoder and encoding/decoding system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106531175A (en) | 2017-03-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| TW580691B (en) | Method and apparatus for interoperability between voice transmission systems during speech inactivity | |
| CN101087319B (en) | A method and device for sending and receiving background noise and silence compression system | |
| US8190440B2 (en) | Sub-band codec with native voice activity detection | |
| CN103187065B (en) | The disposal route of voice data, device and system | |
| TWI332193B (en) | Method and apparatus of processing time-varying signals coding and decoding and computer program product | |
| JP5096582B2 (en) | Noise generating apparatus and method | |
| CN103137133B (en) | Inactive sound modulated parameter estimating method and comfort noise production method and system | |
| WO2015154397A1 (en) | Noise signal processing and generation method, encoder/decoder and encoding/decoding system | |
| CN101246688B (en) | Method, system and device for coding and decoding ambient noise signal | |
| TW200926144A (en) | Spectral noise shaping in audio coding based on spectral dynamics in frequency sub-bands | |
| WO2000046789A1 (en) | Sound presence detector and sound presence/absence detecting method | |
| CN102165699A (en) | Method and apparatus for signal processing using transform-domain log-companding | |
| KR100847391B1 (en) | Method of comfort noise generation for speech communication | |
| CN108231083A (en) | A kind of speech coder code efficiency based on SILK improves method | |
| CN102985968A (en) | Method and device for processing audio signal | |
| CN103680509B (en) | A kind of voice signal discontinuous transmission and ground unrest generation method | |
| CN106531175B (en) | A kind of method that network phone comfort noise generates | |
| CN101483495A (en) | Background noise generation method and noise processing apparatus | |
| JP5415460B2 (en) | Method and means for encoding background noise information | |
| CN101783142A (en) | Transcoding method, device and communication equipment | |
| CN114863942B (en) | Model training method for voice quality conversion, method and device for improving voice quality | |
| JP4437011B2 (en) | Speech encoding device | |
| CN101533639A (en) | Voice signal processing method and device | |
| CN102930872A (en) | Method and device for postprocessing pitch enhancement in broadband speech decoding | |
| PL1756806T3 (en) | Method for quantifying an ultra low-rate speech encoder |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |