Automatic speech segmentation based on alignment with a text-to-speech system
Author(s)
Horák, Petr (URE-Y)
Issue data
Chichester: J.Wiley, 2002
ISBN
0-471-49985-4
Source Title
Improvements in Speech Synthesis / Keller E. ; Bailly G. ; Monaghan A. ; Terken J. ; Huckvale M.
Pages
s. 328-338
Number of pages
11 s.
Language
eng - English
Country
GB - United Kingdom
Keywords
speech processing ; speech synthesis
Subject RIV
BD - Theory of Information
R&D Projects
GV102/96/K087 GA ČR - Czech Science Foundation (CSF)
OC 258.10 GA MŠMT - Ministry of Education, Youth and Sports (MEYS)
CEZ
AV0Z2067918 - URE-Y
Annotation
Automatic phonetic speech segmentation, or the alignment of a known phonetic transcription to a speech signal, is an important tool for many fields of speech reasearch. Most systems for automatic segmentation are based on a trained recognition system. Such systems are typically trained on hidden Markov models of phoneme realizations. An alternative strategy is to use a text-to-speech system to generate a prototype realization of the transcription and to align the synthetic signal with the real one.
