Singing Voice Conversion to Growl-Like Singing Based on Vocal Tract Shape and Glottal Source Characteristics

Shohei MIZOBUCHI  Ryuichi NISIMURA  Toshie MATSUI  Toshio IRINO  Hideki KAWAHARA  

Publication
D - Abstracts of IEICE TRANSACTIONS on Information and Systems (Japanese Edition)   Vol.J99-D   No.3   pp.283-292
Publication Date: 2016/03/01
Online ISSN: 1881-0225
DOI: 
Type of Manuscript: Special Section PAPER (Special Section on Student Research)
Category: 
Keyword: 
growl,  source characteristics,  vocal tract shape,  voice conversion,  TANDEM-STRAIGHT,  

Full Text(in Japanese): PDF(1.1MB)
>>Buy this Article


Summary: 
We have investigated a system to convert usual plain voice to growl-like performance. Relatively high-speed periodic variations in spectral shapes, fundamental frequency trajectories and spectral level enhancement in the 1kHz to 4kHz frequency band were introduced as dominant features of growl-like singing in our previous works. In the previous method, the high-speed periodic variations in spectral shapes designed using a set of Gaussian functions did not take into account of a vocalization mechanism. This study has proposed a method that spectral shape variations can be closely replicated by introducing vocal tract shape variations around supra-laryngeal structure and shape variations in glottal source waveform using the LF-model. In this paper, we have evaluated the proposed method with the previous method by using a pair comparison method. The analysis of the experimental results suggested that the production of singing having the growl-like impression by the proposed method have better than that of the previous method.