Real-Time Robust Voice Activity Detection Using the Upper Envelope Weighted Entropy Measure and the Dual-Rate Adaptive Nonlinear Filter

Abstract

Voice activity detection (VAD) is a vital process in voice communication systems to avoid unnecessary coding and transmission of noise. Most of the existing VAD algorithms continue to suffer high false alarm rates and low sensitivity when the signal-to-noise ratio (SNR) is low, at 0 dB and below. Others are developed to operate in offline mode or are impractical for implementation in actual devices due to high computational complexity. This paper proposes the upper envelope weighted entropy (UEWE) measure as a means to enable high separation of speech and non-speech segments in voice communication. The asymmetric nonlinear filter (ANF) is employed in UEWE to extract the adaptive weight factor that is subsequently used to compensate the noise effect. In addition, this paper also introduces a dual-rate adaptive nonlinear filter (DANF) with high adaptivity to rapid time-varying noise for computation of the decision threshold. Performance comparison with standard and recent VADs shows that the proposed algorithm is superior especially in real-time practical applications.

Robust Audio Content Classification Using Hybrid-Based SMD and Entropy-Based VAD

“…The VAD differs from speech/music discrimination (SMD). VAD discriminates between noise and speech while SMD discriminates between speech and music [1][2][3][4][5][6][7].…”

“…Knowing how to build a management of multimedia information for AV (audio/video) indexing and retrieval is becoming extremely important. In the field of AV indexing and retrieval, the speech/music discrimination (SMD) is a very crucial task for the audio content classification (ACC) system or general audio detection and classification (GADC) [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18]. In recently, the SMD literatures have been presented in different application [19][20][21][22][23][24] and closely related to retrieval of audio content indexing [20].…”

“…To alleviate this problem, the robust spectral entropybased scheme of voice-activity detection (VAD) which distinguishes speech and non-speech segments from the incoming audio signal is combined with the utilized SMD approach as a front-end of the proposed system of ACC application. Especially for the VAD case, the idea of using spectral entropy and other related parameters that monitor spectral variability or flatness has been used for many years [1,3,4]. Our previous research article [56] proved that spectral entropy-based VAD can be successfully applied to a variable noise-level environment.…”