CN1697018A - Method for raising precision of identifying speech by using improved subtractive method of spectrums - Google Patents
Method for raising precision of identifying speech by using improved subtractive method of spectrums Download PDFInfo
- Publication number
- CN1697018A CN1697018A CNA2005100404006A CN200510040400A CN1697018A CN 1697018 A CN1697018 A CN 1697018A CN A2005100404006 A CNA2005100404006 A CN A2005100404006A CN 200510040400 A CN200510040400 A CN 200510040400A CN 1697018 A CN1697018 A CN 1697018A
- Authority
- CN
- China
- Prior art keywords
- spectrums
- speech
- voice
- subtractive method
- precision
- 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
- Machine Translation (AREA)
- Electrically Operated Instructional Devices (AREA)
Abstract
A method using improved spectral phase substraction to raise identification accuracy of pronunciation includes carrying out pronunciation enhancing treatment for pronunciation with noise before pronunciation identification is carried out and applying improved spectral phase substraction for pronunciation enhancing treatment to raise gain of signal to noise ratio.
Description
Technical field
The invention relates to the method that improves precision of identifying speech, specifically, is about before the speech recognition algorithm feature extraction, by voice are carried out enhancement process, improves the phonetic entry signal to noise ratio (S/N ratio), and then improves the method for precision of identifying speech.
Background technology
Speech recognition is the hi-tech that machine changes voice signal into by identification and understanding process corresponding text file or order.As the particular study field, speech recognition is again a cross discipline, and it closely links to each other with numerous subjects such as acoustics, phonetics, linguistics, digital signal processing theory, information theory, computer science.Speech recognition develops into today through four more than ten years, has demonstrated great application prospect.The key of impelling speech recognition technology to be used widely is an accuracy of identification.
The invention provides a kind of by voice being carried out enhancement process, raising phonetic entry signal to noise ratio (S/N ratio), and then the method for raising precision of identifying speech.
Summary of the invention
For above purpose, the invention provides a kind of effectively by voice being carried out enhancement process, raising phonetic entry signal to noise ratio (S/N ratio), and then the method for raising precision of identifying speech.This method comprises:
The voice enhancement process is carried out in phonetic entry to the band noise before carrying out voice recognition processing
The voice enhancement processing method adopts improved subtractive method of spectrums
Change traditional subtractive method of spectrums α=2, β=1 is α=2, the subtractive method of spectrums that β=5 are improved
Improved subtractive method of spectrums snr gain obtains to improve
Find that by the test of speech recognition verification platform the precision of identifying speech of improved subtractive method of spectrums is significantly increased
Description of drawings
In claims of present patent application, pointed out theme of the present invention particularly, and clearly it has been proposed patent protection.Yet with reference to following detailed description and accompanying drawing, relevant structure that can better understand the present invention and implementation method with and purpose, feature and advantage.
Fig. 1 represents to strengthen the speech recognition system block diagram that improves precision of identifying speech, y by voice
(n)Be noisy speech, S
(n) +Be the voice after strengthening, W is the output of recognizer;
Fig. 2 represents to improve spectrum-subtraction block diagram, y
(n)Be noisy speech, s
(n)Be clean speech input, d
(n)Be additive noise, noise power spectrum coefficient lambda n (k), θ
kBe phase place, α, β are parameter, and FFT is the fast fourier conversion, and IFFT is anti-fast fourier conversion, S
(n) +Be the voice after strengthening;
Although herein declarative description certain this feature of the present invention and a kind of implementation method, come for the professional and technical personnel Say, many modifications, replacement, variation and equivalent substitution will occur. Therefore, protection scope of the present invention is wanted by appended right The scope of asking is as the criterion.
Claims (3)
1. method of utilizing improved subtractive method of spectrums to improve precision of identifying speech, this method may further comprise the steps:
---band noise phonetic entry y
(n), comprise two: clean speech input s
(n), additive noise d
(n)
---unvoiced segments in voice is estimated the noise power spectrum coefficient lambda n (k) that obtains
---the spectral coefficient of band noise voice and clean speech is respectively: Y
k, S
k, k=0,1,
---traditional subtractive method of spectrums α=2, β=1 obtains voice s
(n)The spectral amplitude coefficient | S
k|=[| Y
k|
α-β λ n (k)
α]
1/ α
---improve subtractive method of spectrums, set α=2, β=5, the voice S after being enhanced
(n) +The spectral amplitude coefficient | S
k +|
---find to be significantly increased by the test of speech recognition verification platform through the precision of identifying speech of above-mentioned processing.
2. method of utilizing improved subtractive method of spectrums to improve precision of identifying speech, it is characterized in that: subtractive method of spectrums mainly is used in before the speech recognition algorithm feature extraction, and voice are carried out enhancement process, thereby improves the phonetic entry signal to noise ratio (S/N ratio).
3. a kind of according to claim 1 method of utilizing improved subtractive method of spectrums to improve precision of identifying speech is characterized in that: α=2 by improving traditional subtractive method of spectrums and the parameter value of β=1 are α=2 and β=5, improve the speech manual range coefficient.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005100404006A CN100358007C (en) | 2005-06-07 | 2005-06-07 | Method for raising precision of identifying speech by using improved subtractive method of spectrums |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005100404006A CN100358007C (en) | 2005-06-07 | 2005-06-07 | Method for raising precision of identifying speech by using improved subtractive method of spectrums |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1697018A true CN1697018A (en) | 2005-11-16 |
| CN100358007C CN100358007C (en) | 2007-12-26 |
Family
ID=35349715
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2005100404006A Expired - Fee Related CN100358007C (en) | 2005-06-07 | 2005-06-07 | Method for raising precision of identifying speech by using improved subtractive method of spectrums |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100358007C (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104064185A (en) * | 2013-03-18 | 2014-09-24 | 联想(北京)有限公司 | Information processing method and system and electronic device |
| CN106023996A (en) * | 2016-06-12 | 2016-10-12 | 杭州电子科技大学 | Sound identification method based on cross acoustic array broadband wave beam formation |
| CN112309414A (en) * | 2020-07-21 | 2021-02-02 | 东莞市逸音电子科技有限公司 | Active noise reduction method based on audio coding and decoding, earphone and electronic equipment |
| CN112312256A (en) * | 2020-07-30 | 2021-02-02 | 深圳市逸音科技有限公司 | Intelligent active noise reduction earphone based on digital communication |
| CN112312258A (en) * | 2020-09-08 | 2021-02-02 | 深圳市逸音科技有限公司 | Intelligent earphone with hearing protection and hearing compensation |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP3457293B2 (en) * | 2001-06-06 | 2003-10-14 | 三菱電機株式会社 | Noise suppression device and noise suppression method |
| CN1162838C (en) * | 2002-07-12 | 2004-08-18 | 清华大学 | Speech intensifying-characteristic weighing-logrithmic spectrum addition method for anti-noise speech recognization |
-
2005
- 2005-06-07 CN CNB2005100404006A patent/CN100358007C/en not_active Expired - Fee Related
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104064185A (en) * | 2013-03-18 | 2014-09-24 | 联想(北京)有限公司 | Information processing method and system and electronic device |
| CN104064185B (en) * | 2013-03-18 | 2017-06-27 | 联想(北京)有限公司 | Information processing method and system, electronic equipment |
| CN106023996A (en) * | 2016-06-12 | 2016-10-12 | 杭州电子科技大学 | Sound identification method based on cross acoustic array broadband wave beam formation |
| CN106023996B (en) * | 2016-06-12 | 2019-08-27 | 杭州电子科技大学 | Sound recognition methods based on cross acoustic array broad-band EDFA |
| CN112309414A (en) * | 2020-07-21 | 2021-02-02 | 东莞市逸音电子科技有限公司 | Active noise reduction method based on audio coding and decoding, earphone and electronic equipment |
| CN112309414B (en) * | 2020-07-21 | 2024-01-12 | 东莞市逸音电子科技有限公司 | Active noise reduction method based on audio encoding and decoding, earphone and electronic equipment |
| CN112312256A (en) * | 2020-07-30 | 2021-02-02 | 深圳市逸音科技有限公司 | Intelligent active noise reduction earphone based on digital communication |
| CN112312256B (en) * | 2020-07-30 | 2023-08-01 | 深圳市逸音科技有限公司 | Intelligent active noise reduction earphone based on digital communication |
| CN112312258A (en) * | 2020-09-08 | 2021-02-02 | 深圳市逸音科技有限公司 | Intelligent earphone with hearing protection and hearing compensation |
Also Published As
| Publication number | Publication date |
|---|---|
| CN100358007C (en) | 2007-12-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Prasad et al. | Automatic segmentation of continuous speech using minimum phase group delay functions | |
| CN113012720B (en) | Depression detection method by multi-voice feature fusion under spectral subtraction noise reduction | |
| CN109767756B (en) | Sound characteristic extraction algorithm based on dynamic segmentation inverse discrete cosine transform cepstrum coefficient | |
| CN108154879B (en) | Non-specific human voice emotion recognition method based on cepstrum separation signal | |
| CN108564965B (en) | Anti-noise voice recognition system | |
| Sathe-Pathak et al. | Extraction of Pitch and Formants and its Analysis to identify 3 different emotional states of a person | |
| CN101136199A (en) | Voice data processing method and equipment | |
| CN1650349A (en) | On-line parametric histogram normalization for noise robust speech recognition | |
| Chuang et al. | Speaker-Aware Deep Denoising Autoencoder with Embedded Speaker Identity for Speech Enhancement. | |
| CN110853629A (en) | Speech recognition digital method based on deep learning | |
| CN1697018A (en) | Method for raising precision of identifying speech by using improved subtractive method of spectrums | |
| CN106782503A (en) | Automatic speech recognition method based on physiologic information in phonation | |
| CN106297769B (en) | A kind of distinctive feature extracting method applied to languages identification | |
| CN106782599A (en) | The phonetics transfer method of post filtering is exported based on Gaussian process | |
| CN114283822A (en) | Many-to-one voice conversion method based on gamma pass frequency cepstrum coefficient | |
| CN113421584A (en) | Audio noise reduction method and device, computer equipment and storage medium | |
| JP2010102129A (en) | Fundamental frequency extracting method, fundamental frequency extracting device, and program | |
| CN114550741A (en) | Semantic recognition method and system | |
| CN106228984A (en) | Voice recognition information acquisition methods | |
| Alam et al. | Regularized minimum variance distortionless response-based cepstral features for robust continuous speech recognition | |
| Upadhyay et al. | Robust recognition of English speech in noisy environments using frequency warped signal processing | |
| Mahmood | Arabic speaker recognition system based on phoneme fusion | |
| Singh et al. | A comparative study on feature extraction techniques for language identification | |
| CN111048072A (en) | Voiceprint recognition method applied to power enterprises | |
| Saxena et al. | Extricate Features Utilizing Mel Frequency Cepstral Coefficient in Automatic Speech Recognition System |
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 | ||
| C19 | Lapse of patent right due to non-payment of the annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |