Robust speech activity detection in the presence of noise.pdf

Robust Detection of Speech Activity in the Presence of Noise Ruhi Sarikaya, John H.L. Hansen Robust Speech Processing Laboratory Center for Spoken Language Research University of Colorado Boulder, Campus Box 594 (Express Mail: 3215 Marine Street, Room E-265) Boulder, Colorado 80309-0594 303 – 735 –5148 (Phone) 303 – 735 – 5072 (Fax) / John.Hansen@, (email) R RSPL C S L R ICSLP-98: Inter. Conf. On Spoken Language Processing, Sydney, Australia, Nov. 30 – Dec. 4, 1998. R. Sarikaya, J.H.L. Hansen, Robust detection of Speech Activity in the Presence of Noise, ICSLP-98: Inter. Conf. on Spoken Language Processing, vol. 4, pp. 1455-1458, Sydney, Australia, Dec. 1998. ROBUST SPEECH ACTIVITY DETECTION IN THE PRESENCE OF NOISE Ruhi Sarikaya and John H. L. Hansen Robust Speech Processing Laboratory Duke University, Box 90291, Durham, NC 27708-0291 /Research/Speech ruhi@ jhlh@ee.duke ABSTRACT This study presents a new approach for robust speech ac- tivity detection (SAD). Our framework is based on HMM recognition of speech versus silence. We model speech as one of fourteen large phone classes whereas silence is repre- sented as a separate model. Individual test utterances are concatenated to simulate read continuous speech for test- ing. The HMM-based algorithm is compared to both an energy based, as well as speech enhancement based, SAD algorithms for clean, 5 dB and 0 dB SNR levels under white Gaussian noise (WGN), aircraft cockpit noise (AIR) and automobile highway noise (HWY). We found that our algorithm provides lower frame error rates than the other two methods especially for HWY noise. Unlike other stud- ies, we evaluate our algorithm on the core test set of the standard TIMIT database. Hence, results can be used as benchmarks to evaluate future systems. 1. INTRODUCTION Speech activity detection (SAD) is one of the fundamental issues in many speech processing tasks such as continu- ous speech recognition and speech enhancement. Reliable