Title
Hybrid HMM/ANN systems for training independent tasks: experiments on Phonebook and related improvements.
Abstract
In this paper, we evaluate multi-Gaussian HMM systemsand hybrid HMM/ANN systems in the frameworkof task independent training for small size (75 words)and medium size (600 words) vocabularies. To do this,we use the Phonebook database [6] which is particularlywell suited to this kind of experiments since (1) itis a very large telephone database and (2) the size andcontent of the test vocabulary is very flexible. For eachsystem, different HMM topologies are compared to testthe influence...
Year
DOI
Venue
1997
10.1109/ICASSP.1997.598872
ICASSP
Keywords
Field
DocType
related improvements,hybrid hmm,ann systems,independent tasks,gaussian processes,training data,neural nets,topology,speech recognition,automatic speech recognition,telephony,artificial neural networks,learning artificial intelligence,speech,context modeling,hidden markov models,system testing,databases
Computer science,Context model,Artificial intelligence,Natural language processing,Artificial neural network,Pattern recognition,System testing,Speech recognition,Network topology,Lexicon,Telephony,Hidden Markov model,Vocabulary
Conference
ISBN
Citations 
PageRank 
0-8186-7919-0
26
4.13
References 
Authors
5
5
Name
Order
Citations
PageRank
Stephane Dupont132443.20
Herve Bourlard215237.75
Olivier Deroo3559.96
Vincent Fontaine4264.13
Jean-Marc Boite512022.02