Automatic Alignment of Phonetic Transcriptions for Russian

Результат исследований: Научные публикации в периодических изданияхстатья

1 Цитирования (Scopus)


This paper presents automatic alignment of Russian phonetic pronunciations using the information about phonetic nature of speech sounds in the aligned transcription sequences. This approach has been tested on 24 hours of speech data and has shown significant improvement in alignment errors has been obtained in comparison with commonly used Levenstein algorithm: the numbers of error has been reduced from 1.1 % to 0.27 %.
Язык оригиналаанглийский
Страницы (с-по)123-128
ЖурналLecture Notes in Computer Science
Номер выпуска8773
СостояниеОпубликовано - 2014

Fingerprint Подробные сведения о темах исследования «Automatic Alignment of Phonetic Transcriptions for Russian». Вместе они формируют уникальный семантический отпечаток (fingerprint).