9th Annual Conference of the International Speech Communication Association

Brisbane, Australia
September 22-26, 2008

Speech Enhancement Based on Novel Two-Step a priori SNR Estimators

Md. Jahangir Alam (1), Douglas O'Shaughnessy (1), Sid-Ahmed Selouani (2)

(1) Université du Québec, Canada; (2) Université du Moncton, Canada

A widely used method to determine the a priori SNR from noisy speech is the decision directed (DD) approach, but the a priori SNR follows the a posteriori SNR with a delay of one frame in speech frames. As a consequence, the performance of the noise reduction system degrades. In order to overcome this artifact, we propose three computationally simple and efficient two-step methods based on the minimum mean square error (MMSE), the maximum a posteriori (MAP) and the joint MAP criteria for the estimation of the a priori SNR for speech enhancement. The proposed methods avoid the delay problem while keeping the advantages of the DD method. The performance of the proposed a priori SNR estimation methods are evaluated and compared with the conventional DD method by extensive objective quality measures and yield better results than the DD approach-based speech enhancement system.

Full Paper

Bibliographic reference.  Alam, Md. Jahangir / O'Shaughnessy, Douglas / Selouani, Sid-Ahmed (2008): "Speech enhancement based on novel two-step a priori SNR estimators", In INTERSPEECH-2008, 565-568.