Continuous Noise Masking Based Vocoder for Statistical Parametric Speech Synthesis

Mohammed Salah AL-RADHI; Tamás Gábor CSAPÓ; Géza NÉMETH

doi:10.1587/transinf.2019EDP7167

IEICE TRANSACTIONS on Information

Continuous Noise Masking Based Vocoder for Statistical Parametric Speech Synthesis

Mohammed Salah AL-RADHI, Tamás Gábor CSAPÓ, Géza NÉMETH

Full Text Views

0

Cite this

Summary :

In this article, we propose a method called “continuous noise masking (cNM)” that allows eliminating residual buzziness in a continuous vocoder, i.e. of which all parameters are continuous and offers a simple and flexible speech analysis and synthesis system. Traditional parametric vocoders generally show a perceptible deterioration in the quality of the synthesized speech due to different processing algorithms. Furthermore, an inaccurate noise resynthesis (e.g. in breathiness or hoarseness) is also considered to be one of the main underlying causes of performance degradation, leading to noisy transients and temporal discontinuity in the synthesized speech. To overcome these issues, a new cNM is developed based on the phase distortion deviation in order to reduce the perceptual effect of the residual noise, allowing a proper reconstruction of noise characteristics, and model better the creaky voice segments that may happen in natural speech. To this end, the cNM is designed to keep only voice components under a condition of the cNM threshold while discarding others. We evaluate the proposed approach and compare with state-of-the-art vocoders using objective and subjective listening tests. Experimental results show that the proposed method can reduce the effect of residual noise and can reach the quality of other sophisticated approaches like STRAIGHT and log domain pulse model (PML).

Publication: IEICE TRANSACTIONS on Information Vol.E103-D No.5 pp.1099-1107

Publication Date: 2020/05/01

Publicized: 2020/02/10

Online ISSN: 1745-1361

DOI: 10.1587/transinf.2019EDP7167

Type of Manuscript: PAPER

Category: Speech and Hearing

Authors

Mohammed Salah AL-RADHI
  Budapest University of Technology and Economics
Tamás Gábor CSAPÓ
  Budapest University of Technology and Economics,MTA-ELTE Lendület Lingual Articulation Research Group
Géza NÉMETH
  Budapest University of Technology and Economics

Keyword

noise masking, continuous vocoder, speech synthesis, phase distortion, kernel density functions

Cite this

Copy

Mohammed Salah AL-RADHI, Tamás Gábor CSAPÓ, Géza NÉMETH, "Continuous Noise Masking Based Vocoder for Statistical Parametric Speech Synthesis" in IEICE TRANSACTIONS on Information, vol. E103-D, no. 5, pp. 1099-1107, May 2020, doi: 10.1587/transinf.2019EDP7167.
Abstract: In this article, we propose a method called “continuous noise masking (cNM)” that allows eliminating residual buzziness in a continuous vocoder, i.e. of which all parameters are continuous and offers a simple and flexible speech analysis and synthesis system. Traditional parametric vocoders generally show a perceptible deterioration in the quality of the synthesized speech due to different processing algorithms. Furthermore, an inaccurate noise resynthesis (e.g. in breathiness or hoarseness) is also considered to be one of the main underlying causes of performance degradation, leading to noisy transients and temporal discontinuity in the synthesized speech. To overcome these issues, a new cNM is developed based on the phase distortion deviation in order to reduce the perceptual effect of the residual noise, allowing a proper reconstruction of noise characteristics, and model better the creaky voice segments that may happen in natural speech. To this end, the cNM is designed to keep only voice components under a condition of the cNM threshold while discarding others. We evaluate the proposed approach and compare with state-of-the-art vocoders using objective and subjective listening tests. Experimental results show that the proposed method can reduce the effect of residual noise and can reach the quality of other sophisticated approaches like STRAIGHT and log domain pulse model (PML).
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2019EDP7167/_p

Copy

@ARTICLE{e103-d_5_1099,
author={Mohammed Salah AL-RADHI, Tamás Gábor CSAPÓ, Géza NÉMETH, },
journal={IEICE TRANSACTIONS on Information},
title={Continuous Noise Masking Based Vocoder for Statistical Parametric Speech Synthesis},
year={2020},
volume={E103-D},
number={5},
pages={1099-1107},
abstract={In this article, we propose a method called “continuous noise masking (cNM)” that allows eliminating residual buzziness in a continuous vocoder, i.e. of which all parameters are continuous and offers a simple and flexible speech analysis and synthesis system. Traditional parametric vocoders generally show a perceptible deterioration in the quality of the synthesized speech due to different processing algorithms. Furthermore, an inaccurate noise resynthesis (e.g. in breathiness or hoarseness) is also considered to be one of the main underlying causes of performance degradation, leading to noisy transients and temporal discontinuity in the synthesized speech. To overcome these issues, a new cNM is developed based on the phase distortion deviation in order to reduce the perceptual effect of the residual noise, allowing a proper reconstruction of noise characteristics, and model better the creaky voice segments that may happen in natural speech. To this end, the cNM is designed to keep only voice components under a condition of the cNM threshold while discarding others. We evaluate the proposed approach and compare with state-of-the-art vocoders using objective and subjective listening tests. Experimental results show that the proposed method can reduce the effect of residual noise and can reach the quality of other sophisticated approaches like STRAIGHT and log domain pulse model (PML).},
keywords={},
doi={10.1587/transinf.2019EDP7167},
ISSN={1745-1361},
month={May},}

Copy

TY - JOUR
TI - Continuous Noise Masking Based Vocoder for Statistical Parametric Speech Synthesis
T2 - IEICE TRANSACTIONS on Information
SP - 1099
EP - 1107
AU - Mohammed Salah AL-RADHI
AU - Tamás Gábor CSAPÓ
AU - Géza NÉMETH
PY - 2020
DO - 10.1587/transinf.2019EDP7167
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E103-D
IS - 5
JA - IEICE TRANSACTIONS on Information
Y1 - May 2020
AB - In this article, we propose a method called “continuous noise masking (cNM)” that allows eliminating residual buzziness in a continuous vocoder, i.e. of which all parameters are continuous and offers a simple and flexible speech analysis and synthesis system. Traditional parametric vocoders generally show a perceptible deterioration in the quality of the synthesized speech due to different processing algorithms. Furthermore, an inaccurate noise resynthesis (e.g. in breathiness or hoarseness) is also considered to be one of the main underlying causes of performance degradation, leading to noisy transients and temporal discontinuity in the synthesized speech. To overcome these issues, a new cNM is developed based on the phase distortion deviation in order to reduce the perceptual effect of the residual noise, allowing a proper reconstruction of noise characteristics, and model better the creaky voice segments that may happen in natural speech. To this end, the cNM is designed to keep only voice components under a condition of the cNM threshold while discarding others. We evaluate the proposed approach and compare with state-of-the-art vocoders using objective and subjective listening tests. Experimental results show that the proposed method can reduce the effect of residual noise and can reach the quality of other sophisticated approaches like STRAIGHT and log domain pulse model (PML).
ER -

IEICE TRANSACTIONS on Information