CA2195445C - Procede et appareil pour la reconaissance vocale a l'aide d'une liaison a melange partiel et optimise de probabilites - Google Patents

Procede et appareil pour la reconaissance vocale a l'aide d'une liaison a melange partiel et optimise de probabilites Download PDF

Info

Publication number
CA2195445C
CA2195445C CA002195445A CA2195445A CA2195445C CA 2195445 C CA2195445 C CA 2195445C CA 002195445 A CA002195445 A CA 002195445A CA 2195445 A CA2195445 A CA 2195445A CA 2195445 C CA2195445 C CA 2195445C
Authority
CA
Canada
Prior art keywords
speech
cluster
states
probability
codebook
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related
Application number
CA002195445A
Other languages
English (en)
Other versions
CA2195445A1 (fr
Inventor
Vassilios Digalakis
Hy Murveit
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
SRI International Inc
Original Assignee
SRI International Inc
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Priority claimed from US08/276,742 external-priority patent/US5825978A/en
Application filed by SRI International Inc filed Critical SRI International Inc
Publication of CA2195445A1 publication Critical patent/CA2195445A1/fr
Application granted granted Critical
Publication of CA2195445C publication Critical patent/CA2195445C/fr
Anticipated expiration legal-status Critical
Expired - Fee Related legal-status Critical Current

Links

Landscapes

  • Character Discrimination (AREA)
  • Machine Translation (AREA)

Abstract

La présente invention a pour objet un dispositif de reconnaissance vocale qui utilise un procédé pouvant être mis en oeuvre par calculs pour construire un ensemble de modèles de Markov cachés pour la reconnaissance vocale ayant recours à un degré de liaison à mélange optimal et partiel de probabilités. Avec les modèles de Markov cachés partiellement liés, la précision de reconnaissance améliorée d'un vaste corpus de mots de vocabulaire par rapport aux systèmes utilisant des modèles de Markov totalement liés est assurée par un temps système de calcul inférieur au système entièrement non lié. La technique pouvant être mise en oeuvre par calcul comprend les étapes consistant à déterminer un groupe d'états de modèles de Markov cachés qui partagent des composantes de Gauss proches les unes des autres, à développer un livre de codes pour ces groupes, et à recalculer les courbes de Gauss dans le livre de codes pour mieux évaluer les états groupés.
CA002195445A 1994-07-18 1995-07-13 Procede et appareil pour la reconaissance vocale a l'aide d'une liaison a melange partiel et optimise de probabilites Expired - Fee Related CA2195445C (fr)

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
US08/276,742 US5825978A (en) 1994-07-18 1994-07-18 Method and apparatus for speech recognition using optimized partial mixture tying of HMM state functions
US08/276,742 1994-07-18
PCT/US1995/008816 WO1996002912A1 (fr) 1994-07-18 1995-07-13 Procede et appareil pour la reconaissance vocale a l'aide d'une liaison a melange partiel et optimise de probabilites

Publications (2)

Publication Number Publication Date
CA2195445A1 CA2195445A1 (fr) 1996-02-01
CA2195445C true CA2195445C (fr) 2005-11-15

Family

ID=35530880

Family Applications (1)

Application Number Title Priority Date Filing Date
CA002195445A Expired - Fee Related CA2195445C (fr) 1994-07-18 1995-07-13 Procede et appareil pour la reconaissance vocale a l'aide d'une liaison a melange partiel et optimise de probabilites

Country Status (1)

Country Link
CA (1) CA2195445C (fr)

Also Published As

Publication number Publication date
CA2195445A1 (fr) 1996-02-01

Similar Documents

Publication Publication Date Title
US5825978A (en) Method and apparatus for speech recognition using optimized partial mixture tying of HMM state functions
Digalakis et al. Genones: Generalized mixture tying in continuous hidden Markov model-based speech recognizers
US7590537B2 (en) Speaker clustering and adaptation method based on the HMM model variation information and its apparatus for speech recognition
JP2871561B2 (ja) 不特定話者モデル生成装置及び音声認識装置
US5165007A (en) Feneme-based Markov models for words
US5729656A (en) Reduction of search space in speech recognition using phone boundaries and phone ranking
US5822729A (en) Feature-based speech recognizer having probabilistic linguistic processor providing word matching based on the entire space of feature vectors
EP0804721A2 (fr) Procede et appareil de reconnaissance vocale adaptee a un locuteur individuel
WO1996022514A9 (fr) Procede et appareil de reconnaissance vocale adaptee a un locuteur individuel
US7587321B2 (en) Method, apparatus, and system for building context dependent models for a large vocabulary continuous speech recognition (LVCSR) system
CN102237082B (zh) 语音识别系统的自适应方法
Euler et al. Statistical segmentation and word modeling techniques in isolated word recognition
Bacchiani et al. Design of a speech recognition system based on acoustically derived segmental units
JP3589044B2 (ja) 話者適応化装置
Walter et al. An evaluation of unsupervised acoustic model training for a dysarthric speech interface
Ko et al. Eigentriphones for context-dependent acoustic modeling
Kannadaguli et al. Phoneme modeling for speech recognition in Kannada using Hidden Markov Model
Torbati et al. A nonparametric Bayesian approach for spoken term detection by example query
JP3216565B2 (ja) 音声モデルの話者適応化方法及びその方法を用いた音声認識方法及びその方法を記録した記録媒体
EP0238693B1 (fr) Procédé et dispositif de reconnaissance de la parole utilisant des représentations statistiques des mots
Han et al. Trajectory clustering for solving the trajectory folding problem in automatic speech recognition
Matsui et al. Smoothed N-best-based speaker adaptation for speech recognition
Digalakis et al. Continuous Speech Dictation on ARPA's North American Business News Domain
De Mori et al. Search and learning strategies for improving hidden Markov models
Digalakis et al. High-accuracy large-vocabulary speech recognition using mixture tying and consistency modeling

Legal Events

Date Code Title Description
EEER Examination request
MKLA Lapsed

Effective date: 20150713