Method And System Of Estimating Clean Speech Parameters From Noisy Speech Parameters

Tata Consultancy Services Limited
Granted App No 201621009058 Patent 462124 Electronics

Patent information

IP India · live record
Application number
201621009058
Filing date
15 March 2016
Status
Granted
Legal status
Inforce
Publication number
46/2017
Publication type
INA
Invention field
Electronics
Patent number
462124
Grant date
26 October 2023
Next renewal due
15 March 2026Overdue

Abstract

A method and system is provided for estimating clean speech parameters from noisy speech parameters. The method is performed by acquiring speech signals, estimating noise from the acquired speech signals, computing speech features from the acquired speech signals, estimating model parameters from the computed speech features and estimating clean parameters from the estimated noise and the estimated model parameters.

Applicants

1
India
TATA CONSULTANCY SERVICES LIMITED Nirmal Building, 9th Floor, Nariman Point, Mumbai-400021, Maharashtra, India

Inventors

2
India
PANDA, AshishTata Consultancy Services Limited Desk No. 5G2, Innovation Lab Yantra Park -(STPI), 2nd Pokharan Road, Subash Nagar Unit No. 6 Thane-400601, Maharashtra, India
India
KOPPARAPU, Sunil KumarTata Consultancy Services Limited Desk No. 5G2, Innovation Lab Yantra Park -(STPI), 2nd Pokharan Road, Subash Nagar Unit No. 6 Thane-400601, Maharashtra, India

Claims

9
1 A method of estimating clean speech parameters from noisy speech parameters, said method comprising processor implemented steps of: acquisition of speech signals using a speech acquisition module (202); estimation of noise from said acquired speech signals using a noise estimation module (204); computation of speech features from said acquired speech signals using a feature extraction module (206); estimation of model parameters from the said computed speech features using a parameter estimation module (208); estimation of clean parameters from said estimated noise and said estimated model parameters using a clean parameter estimation module (210);
2 The method as claimed in claim 1, wherein said speech acquisition module (202) further converts the said acquired speech signals from analog to digital waveforms.
3 The method as claimed in claim 1, wherein said estimation of noise using the noise estimation module is performed during training phase.
4 The method as claimed in claim 1, wherein said estimation of noise can further be performed through non-speech frames of said acquired speech signals.
5 The method as claimed in claim 1, wherein Mel-Frequency Cepstral Coefficients are used as said speech features in the feature extraction module.
6 The method as claimed in claim 1, wherein said estimated noise and said estimated model parameters are first converted into their spectral domain representations.
7 The method as claimed in claim 1, wherein the estimated clean parameters are in their spectral domain representation.
8 The method as claimed in Claim 7, wherein the estimated clean parameters are converted from their spectral domain representation to feature domain representation.
9 A system of estimating clean speech parameters from noisy speech parameters, said system comprising: a processor; a data bus coupled to said processor; and a computer-usable medium embodying computer code, said computer-usable medium being coupled to said data bus, said computer program code comprising instructions executable by said processor and configured for operating a speech acquisition module (202) adapted for acquiring speech signals; a noise estimation module (204) adapted for estimating noise from said acquired speech signals; a feature extraction module (206) adapted for computing speech features from said acquired speech signals; a parameter estimation module (208) adapted for estimating model parameters from the said computed speech features; a clean parameter estimation module (210) adapted for estimating clean parameters from said estimated noise and said estimated model parameters. , Description:As Attached

Specification

Show full textCollapse

Claims:1. A method of estimating clean speech parameters from noisy speech parameters, said method comprising processor implemented steps of:

acquisition of speech signals using a speech acquisition module (202); estimation of noise from said acquired speech signals using a noise estimation module (204); computation of speech features from said acquired speech signals using a feature extraction module (206); estimation of model parameters from the said computed speech features using a parameter estimation module (208); estimation of clean parameters from said estimated noise and said estimated model parameters using a clean parameter estimation module (210);

2. The method as claimed in claim 1, wherein said speech acquisition module (202) further converts the said acquired speech signals from analog to digital waveforms.

3. The method as claimed in claim 1, wherein said estimation of noise using the noise estimation module is performed during training phase.

4. The method as claimed in claim 1, wherein said estimation of noise can further be performed through non-speech frames of said acquired speech signals.

5. The method as claimed in claim 1, wherein Mel-Frequency Cepstral Coefficients are used as said speech features in the feature extraction module.

6. The method as claimed in claim 1, wherein said estimated noise and said estimated model parameters are first converted into their spectral domain representations.

7. The method as claimed in claim 1, wherein the estimated clean parameters are in their spectral domain representation.

8. The method as claimed in Claim 7, wherein the estimated clean parameters are converted from their spectral domain representation to feature domain representation.

9. A system of estimating clean speech parameters from noisy speech parameters, said system comprising:

a processor; a data bus coupled to said processor; and a computer-usable medium embodying computer code, said computer-usable medium being coupled to said data bus, said computer program code comprising instructions executable by said processor and configured for operating a speech acquisition module (202) adapted for acquiring speech signals; a noise estimation module (204) adapted for estimating noise from said acquired speech signals; a feature extraction module (206) adapted for computing speech features from said acquired speech signals; a parameter estimation module (208) adapted for estimating model parameters from the said computed speech features; a clean parameter estimation module (210) adapted for estimating clean parameters from said estimated noise and said estimated model parameters.

, Description:As Attached

Documents

Sign in to open
Application documents
1 Form 5 [15-03-2016(online)].pdf 15 Mar 2016
2 Form 3 [15-03-2016(online)].pdf 15 Mar 2016
3 Form 18 [15-03-2016(online)].pdf 15 Mar 2016
4 Drawing [15-03-2016(online)].pdf 15 Mar 2016
5 Description(Complete) [15-03-2016(online)].pdf 15 Mar 2016
6 Form 26 [04-07-2016(online)].pdf 04 Jul 2016
7 201621009058-POWER OF ATTORNEY-(07-07-2016).pdf 07 Jul 2016
8 201621009058-CORRESPONDENCE-(07-07-2016).pdf 07 Jul 2016
9 Other Patent Document [31-08-2016(online)].pdf 31 Aug 2016
10 REQUEST FOR CERTIFIED COPY [06-02-2017(online)].pdf 06 Feb 2017
11 Form 3 [08-03-2017(online)].pdf 08 Mar 2017
12 Request For Certified Copy-Online.pdf 11 Aug 2018
13 Abstract.jpg 11 Aug 2018
14 201621009058-Form 1-060916.pdf 11 Aug 2018
15 201621009058-Correspondence-060916.pdf 11 Aug 2018
16 201621009058-CORRESPONDENCE(IPO)-(CERTIFIED)-(14-2-2017).pdf 11 Aug 2018
17 201621009058-FER.pdf 28 Aug 2019
18 201621009058-Information under section 8(2) [11-02-2020(online)].pdf 11 Feb 2020
19 201621009058-FORM 3 [11-02-2020(online)].pdf 11 Feb 2020
20 201621009058-OTHERS [25-02-2020(online)].pdf 25 Feb 2020
21 201621009058-FER_SER_REPLY [25-02-2020(online)].pdf 25 Feb 2020
22 201621009058-DRAWING [25-02-2020(online)].pdf 25 Feb 2020
23 201621009058-CLAIMS [25-02-2020(online)].pdf 25 Feb 2020
24 201621009058-ABSTRACT [25-02-2020(online)].pdf 25 Feb 2020
25 201621009058-PatentCertificate26-10-2023.pdf 26 Oct 2023
26 201621009058-IntimationOfGrant26-10-2023.pdf 26 Oct 2023
Search strategy
12019-08-0512-32-51_27-08-2019.pdf

ERegister / renewals

Inforce
3rd Renewed 06 Nov 2023 Period 15 Mar 2018 – 15 Mar 2019 · CBR 45368 · Certificate 46403 ₹4,000
4th Renewed 06 Nov 2023 Period 15 Mar 2019 – 15 Mar 2020 · CBR 45368 · Certificate 46404 ₹4,000
5th Renewed 06 Nov 2023 Period 15 Mar 2020 – 15 Mar 2021 · CBR 45368 · Certificate 46405 ₹4,000
6th Renewed 06 Nov 2023 Period 15 Mar 2021 – 15 Mar 2022 · CBR 45368 · Certificate 46406 ₹4,000
7th Renewed 06 Nov 2023 Period 15 Mar 2022 – 15 Mar 2023 · CBR 45368 · Certificate 46407 ₹12,000
8th Renewed 06 Nov 2023 Period 15 Mar 2023 – 15 Mar 2024 · CBR 45368 · Certificate 46408 ₹12,000
9th Renewed 06 Nov 2023 Period 15 Mar 2024 – 15 Mar 2025 · CBR 45368 · Certificate 46409 ₹12,000
10th Renewed 12 Mar 2025 Period 15 Mar 2025 – 15 Mar 2026 · CBR 9582 · Certificate 11491 ₹12,000