On the use of spectral transformation for speaker adaptation in HMM based isolated-word speech recognition

Choi, Hyun-Jun; King, Robin

doi:10.1016/0167-6393(95)00019-k

Cited by 7 publications

(7 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Therefore MMSE estimation will lose much second order statistical information. That is why Choi and King reported that canonical correlation analysis (CCA) estimation outperformed MMSE in speaker adaptation using spectral transformation, because the variance of each component of the transformed vectors is also explicitly considered in the CCA method, which is not the case in MMSE [7]. On the other hand, the relationship between source and target spectral feature space is not linear in fact.…”

Section: ⅰ Intruductionmentioning

confidence: 94%

An Approach to Voice Conversion Based on Non-Linear Canonical Correlation Analysis

Jian¹

2009

2009 5th International Conference on Wireless Communications, Networking and Mobile Computing

View full text Add to dashboard Cite

Voice conversion algorithm aims to provide high level of similarity to the target voice with an acceptable level of quality. The main object of this paper was to build a nonlinear relationship between the parameters for the acoustical features of source and target speaker using non-linear canonical correlation analysis (NLCCA) based on jointed Gaussian mixture model. Speaker individuality transformation was achieved mainly by altering vocal tract characteristics represented by line spectral frequencies (LSF). To obtain the transformed speech sounded more like the target voices, prosody modification is involved through residual prediction. Both objective and subjective evaluations were conducted. The experimental results demonstrated that our proposed algorithm was effective and outperformed the conventional conversion method utilized by the minimum mean square error (MMSE) estimation.

show abstract

Section: ⅰ Intruductionmentioning

confidence: 94%

An Approach to Voice Conversion Based on Non-Linear Canonical Correlation Analysis

Jian¹

2009

2009 5th International Conference on Wireless Communications, Networking and Mobile Computing

View full text Add to dashboard Cite

show abstract

“…Therefore, MMSE estimation will lose much second order statistical information. That is why Choi and King reported that Canonical Correlation Analysis (CCA) estimation outperformed MMSE in speaker adaptation using spectral transformation, because the variance of each component of the transformed vectors is also explicitly considered in the CCA method, which is not the case in MMSE [10] . On the other hand, the relationship between source and target spectral feature space is not linear actually.…”

Section: Introductionmentioning

confidence: 98%

On using non-linear canonical correlation analysis for voice conversion based on Gaussian mixture model

Jian

Yang

2010

J. Electron.(China)

View full text Add to dashboard Cite

Voice conversion algorithm aims to provide high level of similarity to the target voice with an acceptable level of quality. The main object of this paper was to build a nonlinear relationship between the parameters for the acoustical features of source and target speaker using Non-Linear Canonical Correlation Analysis (NLCCA) based on jointed Gaussian mixture model. Speaker individuality transformation was achieved mainly by altering vocal tract characteristics represented by Line Spectral Frequencies (LSF). To obtain the transformed speech which sounded more like the target voices, prosody modification is involved through residual prediction. Both objective and subjective evaluations were conducted. The experimental results demonstrated that our proposed algorithm was effective and outperformed the conventional conversion method utilized by the Minimum Mean Square Error (MMSE) estimation.

show abstract

“…So the LPC spectral feature contains almost all the second order statistical information in speech if the LPC analysis is performed perfectly. In contrast to the MMSE, Choi and King reported that Canonical Correlation Analysis (CCA) estimation provided better performance by using spectral transformation for speaker adaptation, because the variance of each component of the transformed vectors is also explicitly considered in the CCA method [15] . In this paper, a voice conversion algorithm is proposed, in which CCA is used to estimate the spectral mapping function.…”

Section: Introductionmentioning

confidence: 99%

“…In recent years, stochastic a proaches with continuous map- 1 [2,10,13] . But the MMSE estimation does not consider the variance of each component of the transformed vectors [15] . So it will lose much second order statistical information.…”

Section: Introductionmentioning

confidence: 99%

A novel algorithm for voice conversion using Canonical Correlation Analysis

Jian

Yang

2008

J. Electron.(China)

View full text Add to dashboard Cite

A novel algorithm for voice conversion is proposed in this paper. The mapping function of spectral vectors of the source and target speakers is calculated by the Canonical Correlation Analysis (CCA) estimation based on Gaussian mixture models. Since the spectral envelope feature remains a majority of second order statistical information contained in speech after Linear Prediction Coding (LPC) analysis, the CCA method is more suitable for spectral conversion than Minimum Mean Square Error (MMSE) because CCA explicitly considers the variance of each component of the spectral vectors during conversion procedure. Both objective evaluations and subjective listening tests are conducted. The experimental results demonstrate that the proposed scheme can achieve better performance than the previous method which uses MMSE estimation criterion.

show abstract

On the use of spectral transformation for speaker adaptation in HMM based isolated-word speech recognition

Cited by 7 publications

References 14 publications

An Approach to Voice Conversion Based on Non-Linear Canonical Correlation Analysis

An Approach to Voice Conversion Based on Non-Linear Canonical Correlation Analysis

On using non-linear canonical correlation analysis for voice conversion based on Gaussian mixture model

A novel algorithm for voice conversion using Canonical Correlation Analysis

Contact Info

Product

Resources

About