BibTeX
@MISC{_asegment-based,
author = {},
title = {A SEGMENT-BASED PROBABILISTIC GENERATIVE MODEL OF SPEECH},
year = {}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
ABSTRACT We present a purely time domain approach to speech pro-cessing which identifies waveform samples at the boundaries between glottal pulse periods (in voiced speech) orat the boundaries of unvoiced segments. An efficient algorithm for inferring these boundaries and estimating theaverage spectra of voiced and unvoiced regions is derived from a simple probabilistic generative model. Competitiveresults are presented on pitch tracking, voiced/unvoiced detection and timescale modification; all these tasks and sev-eral others can be performed using the single segmentation provided by inference in the model.
Keyphrases
segment-based probabilistic generative model speech single segmentation purely time domain approach efficient algorithm theaverage spectrum unvoiced segment glottal pulse period simple probabilistic generative model waveform sample sev-eral others pitch tracking unvoiced detection unvoiced region timescale modification
