We propose a new paradigm for aligning a phoneme sequence of a speech utterance with its acoustical signal counterpart. In contrast to common HMM-based approaches, our method employs a discriminative learning procedure in which the learning phase is tightly coupled with the alignment task at hand. The alignment function we devise is based on mapping the input acoustic-symbolic representations of the speech utterance along with the target alignment into an abstract vector space. We suggest a specific mapping into the abstract vector-space which utilizes standard speech features (e.g. spectral distances) as well as confidence outputs of a framewise phoneme classifier. Building on techniques used for large margin methods for predicting whole sequences, our alignment function distills to a classifier in the abstract vectorspace which separates correct alignments from incorrect ones. We describe a simple iterative algorithm for learning the alignment function and discuss its formal properties. Experiments with the TIMIT corpus show that our method outperforms the current state-of-the-art approaches.
Cite as: Keshet, J., Shalev-Shwartz, S., Singer, Y., Chazan, D. (2005) Phoneme alignment based on discriminative learning. Proc. Interspeech 2005, 2961-2964, doi: 10.21437/Interspeech.2005-129
@inproceedings{keshet05_interspeech,
author={Joseph Keshet and Shai Shalev-Shwartz and Yoram Singer and Dan Chazan},
title={{Phoneme alignment based on discriminative learning}},
year=2005,
booktitle={Proc. Interspeech 2005},
pages={2961--2964},
doi={10.21437/Interspeech.2005-129}
}