4th International Conference on Spoken Language Processing

Philadelphia, PA, USA
October 3-6, 1996

Duration Modeling with Expanded HMM Applied to Speech Recognition

Antonio Bonafonte, Josep Vidal, Albino Nogueiras

Universitat Politècnica de Catalunya, c/Gran Capità s/n, Barcelona, Spain

In this paper, the occupancy of the HMM states is modeled by means of a Markov chain. A linear estimator is introduced to compute the probabilities of the Markov chain. The distribution functions (DF) represents accurately the observed data. Representing the DF as a Markov chain allows the use of standard HMM recognizers. The increase of complexity is negligible in training and strongly limited during recognition. Experiments performed on acoustic-phonetic decoding shows how the phone recognition rate increases from 60.6 to 61.1. Furthermore, on a task of database inquires, where phones are used as subword units, the correct word rate increases from 88.2 to 88.4.

Full Paper

Bibliographic reference.  Bonafonte, Antonio / Vidal, Josep / Nogueiras, Albino (1996): "Duration modeling with expanded HMM applied to speech recognition", In ICSLP-1996, 1097-1100.