Skip to Main content Skip to Navigation
Master thesis

Suivi de partition pour l'alignement de la voix chantée

Rong Gong 1, 2
1 MuTant - Synchronous Realtime Processing and Programming of Music Signals
IRCAM - Institut de Recherche et Coordination Acoustique/Musique, Inria Paris-Rocquencourt, UPMC - Université Pierre et Marie Curie - Paris 6, CNRS - Centre National de la Recherche Scientifique
2 Analyse et synthèse sonores;
STMS - Sciences et Technologies de la Musique et du Son, IRCAM - Institut de Recherche et Coordination Acoustique/Musique
Abstract : The pitch of singing voice is relatively less accurate and less stable than that of musical instruments and there are many variabilities, such as vibrato, intonation, etc. The music/score alignment system AnteScofo is based on the pitch information which adapts well to instrument music, but raises reliability problems for singing voice. This internship has two objectives, the first is to integrate the phonetic information into the system alignment to achieve the singing voice/text alignment and the second is to merge this information with that of pitch to make the system more robust. To accomplish the first objective, we integrate the vowel templates into the system, and conduct a preliminary experiment to prove that the spectral envelope gene- rated by the method "True envelope" is sufficiently distinctive for the vowels. To accomplish the second objective, two information fusion methods are proposed and among those the arith- metic averaging of the observation probabilities is proven to be the most robust by the evaluation.
Document type :
Master thesis
Complete list of metadata

https://hal.inria.fr/hal-01066603
Contributor : Rong Gong <>
Submitted on : Sunday, September 21, 2014 - 11:50:13 PM
Last modification on : Tuesday, July 13, 2021 - 2:17:12 PM
Long-term archiving on: : Friday, April 14, 2017 - 12:58:22 PM

File

rapport.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01066603, version 1

Citation

Rong Gong. Suivi de partition pour l'alignement de la voix chantée. Intelligence artificielle [cs.AI]. 2014. ⟨hal-01066603⟩

Share

Metrics

Record views

1788

Files downloads

837