An MELP Vocoder Based on UVS and MVF

Tangle Lu; Xiaoqun Zhao

Machine Learning and Intelligent Communications. First International Conference, MLICOM 2016, Shanghai, China, August 27-28, 2016, Revised Selected Papers

Research Article

An MELP Vocoder Based on UVS and MVF

Download

319 downloads

Cite: BibTeX Plain Text

@INPROCEEDINGS{10.1007/978-3-319-52730-7_5,
    author={Tangle Lu and Xiaoqun Zhao},
    title={An MELP Vocoder Based on UVS and MVF},
    proceedings={Machine Learning and Intelligent Communications. First International Conference, MLICOM 2016, Shanghai, China, August 27-28, 2016, Revised Selected Papers},
    proceedings_a={MLICOM},
    year={2017},
    month={2},
    keywords={Signal processing MELP UVS MVF Speech evaluation},
    doi={10.1007/978-3-319-52730-7_5}
}

Tangle Lu
Xiaoqun Zhao
Year: 2017
An MELP Vocoder Based on UVS and MVF
MLICOM
Springer
DOI: 10.1007/978-3-319-52730-7_5

Tangle Lu¹^,*, Xiaoqun Zhao¹^,*

1: Tongji University Shanghai

*Contact email: lutangle@tongji.edu.cn, zhao_xiaoqun@tongji.edu.cn

Abstract

Mixed excitation linear prediction (MELP) vocoder is generally used in low bit-rate vocoder, whose target now focuses on overall coding scheme, decrease of coding rate and improvement of robustness. Unvoiced/voiced/silence detective algorithm (UVS) possesses certain robustness and anti-noise property, while voiced excitation model based on maximum voicing frequency algorithm (MVF) is closer to the original speech characteristics. In this paper, the original excitation model of MELP vocoder is replaced and UVS is joined so that an improved 2.4 kbps coding rate vocoder is accomplished. Compared with MELP of federal standards, the improved vocoder owns better synthetic speech quality and robustness.

Keywords: Signal processing MELP UVS MVF Speech evaluation

Published: 2017-02-24
Appears in: SpringerLink

: http://dx.doi.org/10.1007/978-3-319-52730-7_5