Text Input Using Speaker-Adaptive Connected Syllable Recognition

Auteurs-es

  • Yoichi Takebayashi
  • Hiroyuki Tsuboi
  • Shouichi Hirai
  • Hiroshi Matsura
  • Tsuneo Nitta

Résumé

This paper describes a speech recognition system for large vocabulary text input. The system recognizes connected Japanese syllables by both continuous pattern matching and speaker-adaptation based on the Multiple Similarity (MS) method. The recognition algorithm consists of syllabic boundary detection, vowel and consonant recognition and lexical verification. The reference pattern vectors adapt to each speaker by K-L expansion through covariance matrix modification. Recognition experiments on a 17,877 word Japanese vocabulary showed 92.6% accuracy for 10 male 4,400 phrase utterances.

Fichiers supplémentaires

Publié-e

2022-12-03

Comment citer

1.
Takebayashi Y, Tsuboi H, Hirai S, Matsura H, Nitta T. Text Input Using Speaker-Adaptive Connected Syllable Recognition. Canadian Acoustics [Internet]. 3 déc. 2022 [cité 14 oct. 2024];14(3 bis):105-6. Disponible à: https://jcaa.caa-aca.ca/index.php/jcaa/article/view/3548

Numéro

Rubrique

Actes du congrès de la Semaine canadienne d'acoustique