Abstract

In this paper we present a general approach to identifying non-linguistic speech features from the recorded signal using phone-based acoustic likelihoods. The basic idea is to process the unknown speech signal by feature-specific phone model sets in parallel, and to hypothesize the feature value associated with the model set having the highest likelihood. This technique is shown to be effective for text-independent gender, speaker and language identification. Text-independent speaker identification accuracies of 98·8% on TIMIT (168 speakers) and 99·2% on BREF (65 speakers), were obtained with one utterance per speaker, and 100% with two utterances for both corpora. Experiments in which speaker-specific models were estimated without using the phonetic transcriptions for the TIMIT speakers had the same identification accuracies as those obtained with the use of the transcriptions. French/English language identification is better than 99% with 2 s of read, laboratory speech. For spontaneous telephone speech from the OGI corpus, the language can be identified as French or English with 82% accuracy with 10 s of speech. The ten language identification rate using the OGI corpus was 59·7% with 10 s of signal.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call