Abstract : This article analyzes the phonetic decoding performance obtained with different choices of linguistic units. The context is to later use such an approach as a support for helping communication with deaf people, and to run it on an embedded decoder on a portable terminal, which introduces constrains on the model size. As a first step, this paper compares the performance of various approaches on the ESTER2 and ETAPE speech corpora. Two baseline systems are considered, one relying on a large vocabulary speech recognizer, and another one relying on a phonetic n-gram language model. The third model which relies on a syllable-based lexicon and a trigram language model, provides a good tradeoff between model size and phonetic decoding performance. The phone error rate is only 4% worse (absolute) than the phone error rate obtained with the large vocabulary recognizer, and much better than the phone error rate obtained with the phone n-gram language model. Phone error rates are then analyzed with respect to SNR and speaking rate.
