An adaptive soft voice activity detector for automatic speech recognition system

In this paper, a novel Adaptive Voice Activity Detection (VAD) algorithm is proposed, which is successfully implemented in a MFCC based speech recognition system. The proposed adaptive VAD describes a novel scheme of detecting speech presence/absence by tracking the higher portion of speech power spectrum. Both clean and noisy speech is used to show the effectiveness of the proposed VAD algorithm. Besides, speech recognition experiments are also conducted. Evaluation tests are carried out using a standard English database, AURORA2. Experimental results show that the proposed algorithm manages to achieve very impressive improvements. Experimental results show that the proposed algorithm manages to achieve significant improvements.