Incorporating Knowledge Sources into Statistical Speech Recognition

Wolfgang Minker Satoshi Nakamura Konstantin Markov Sakriani Sakti

ISBN 10: 0387858296 / ISBN 13: 9780387858296

Editore: Springer, 2009

Riguardo questo articolo

pp. 196 1st Edition. Codice articolo 26472753

Dati bibliografici

Titolo: Incorporating Knowledge Sources into ...

Casa editrice: Springer

Data di pubblicazione: 2009

Legatura: Rilegato

Condizione: Used

Descrizione articolo

Riassunto:

Incorporating Knowledge Sources into Statistical Speech Recognition addresses the problem of developing efficient automatic speech recognition (ASR) systems, which maintain a balance between utilizing a wide knowledge of speech variability, while keeping the training / recognition effort feasible and improving speech recognition performance. The book provides an efficient general framework to incorporate additional knowledge sources into state-of-the-art statistical ASR systems. It can be applied to many existing ASR problems with their respective model-based likelihood functions in flexible ways.

Dalla quarta di copertina:

Incorporating Knowledge Sources into Statistical Speech Recognition offers solutions for enhancing the robustness of a statistical automatic speech recognition (ASR) system by incorporating various additional knowledge sources while keeping the training and recognition effort feasible.

The authors provide an efficient general framework for incorporating knowledge sources into state-of-the-art statistical ASR systems. This framework, which is called GFIKS (graphical framework to incorporate additional knowledge sources), was designed by utilizing the concept of the Bayesian network (BN) framework. This framework allows probabilistic relationships among different information sources to be learned, various kinds of knowledge sources to be incorporated, and a probabilistic function of the model to be formulated.

Incorporating Knowledge Sources into Statistical Speech Recognition demonstrates how the statistical speech recognition system may incorporate additional information sources by utilizing GFIKS at different levels of ASR. The incorporation of various knowledge sources, including background noises, accent, gender and wide phonetic knowledge information, in modeling is discussed theoretically and analyzed experimentally.