Self-organized language modeling for speech recognition
Readings in speech recognition
Prosody in Speech Understanding Systems
Prosody in Speech Understanding Systems
Verbmobil: The Combination of Deep and Shallow Processing for Spontaneous Speech Translation
ICASSP '97 Proceedings of the 1997 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP '97) -Volume 1 - Volume 1
A new class of fenonic Markov word models for large vocabulary continuous speech recognition
ICASSP '91 Proceedings of the Acoustics, Speech, and Signal Processing, 1991. ICASSP-91., 1991 International Conference
| Hi-index | 0.00 |
Successful detection of the position of prosodic phrase boundaries is useful for the rescoring of the sentence hypotheses in a speech recognition system. In addition, knowledge about prosodic boundaries may be used in a speech understanding system for disambiguation. In this paper, a segment oriented approach to prosodic boundary detection is presented. In contrast to word oriented methods (e.g. [6]), it has the advance to be independent of the spoken word chain. This makes it possible to use the knowledge about the boundary positions to reduce search space during word recognition. We have evaluated several different boundary detectors. For the two class problem 'boundary vs. no-boundary' we achieved an average recognition rate of 77% and an overall recognition rate up to 92 %. On the spoken phoneme chain 83% average recognition rate (total 92 %) is possible.