Towards bottom-up continuous phone recognition
Authors:
- 1 January 2007
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 566-569
Abstract
We present a novel approach to designing bottom-up automatic speech recognition (ASR) systems. The key component of the proposed approach is a bank of articulatory attribute detectors implemented using a set of feed-forward artificial neural networks (ANNs). Each detector computes a score describing an activation level of the specified speech attributes that the current frame exhibits. These cues are first combined by an event merger that provides some evidence about the presence of a higher level feature which is then verified by an evidence verifier to produce hypotheses at the phone or word level. We evaluate several configurations of our proposed system on a continuous phone recognition task. Experimental results on the TIMIT database show that the system achieves a phone error rate of 25% which is superior to results obtained with either hidden Markov model (HMM) or conditional random field (CRF) based recognizers. We believe the system's inherent flexibility and the ease of adding new detectors may provide further improvements.Keywords
This publication has 5 references indexed in Scilit:
- Further Experiments with Detector-Based Conditional Random Fields in Phonetic Recognition2007
- A study on lattice rescoring with knowledge scores for automatic speech recognition2006
- Hierarchical Structures of Neural Networks for Phoneme Recognition2006
- On designing and evaluating speech event detectors2005
- Temporal patterns (TRAPs) in ASR of noisy speech1999