Efficient voice activity detection algorithm using long-term spectral flatness measure

Abstract

This paper proposes a novel and robust voice activity detection (VAD) algorithm utilizing long-term spectral flatness measure (LSFM) which is capable of working at 10 dB and lower signal-to-noise ratios(SNRs). This new LSFM-based VAD improves speech detection robustness in various noisy environments by employing a low-variance spectrum estimate and an adaptive threshold. The discriminative power of the new LSFM feature is shown by conducting an analysis of the speech/non-speech LSFM distributions. The proposed algorithm was evaluated under 12 types of noises (11 from NOISEX-92 and speech-shaped noise) and five types of SNR in core TIMIT test corpus. Comparisons with three modern standardized algorithms (ETSI adaptive multi-rate (AMR) options AMR1 and AMR2 and ITU-T G.729) demonstrate that our proposed LSFM-based VAD scheme achieved the best average accuracy rate. A long-term signal variability (LTSV)-based VAD scheme is also compared with our proposed method. The results show that our proposed algorithm outperforms the LTSV-based VAD scheme for most of the noises considered including difficult noises like machine gun noise and speech babble noise.

Voice Activity Detection Using Fuzzy Entropy and Support Vector Machine

“…In subjective evaluation, a human listener evaluates for VAD errors, whereas, numerical computations are carried out for objective evaluation. However, subjective evaluation alone is insufficient to examine the VAD performance, because listening tests like ABC fail to consider the effects of false alarm [32,34,35]. Hence numerical computations through objective evaluation help in reporting the performance of the proposed VAD algorithm.…”

Robust Voice Activity Detection Algorithm based on Long Term Dominant Frequency and Spectral Flatness Measure

“…Dominant frequency component D x (m) of each frame is calculated using steps described in section II-A and spectral envelopes are also estimated. Then we follow the same procedure for LSFM feature L x (m) computation as stated in Yanna Ma et al [15]. The power spectrum of the segmented signal is estimated using Welch-Bartlett method since it is better than periodogram [22].…”

“…Proper selection of this two parameter will increase discriminating power of speech and non-speech and hence a better VAD. We evaluate our proposed algorithm following the method mentioned in [15] using the Edinburgh corpus database and NOISEX92 database (in section IV). Experimentally we also found that the values for R = 30 and M = 10 are same as in [15].…”

“…LTSV shows great improvement in both stationary and non-stationary noise conditions, but its discrimination power drops when SNR is higher than 5dB. Moreover, Yanna ma et al [15] propose long term spectral flatness measure (LSFM) based VAD, which employs a low-variance spectrum estimate and an adaptive threshold. LSFM-based VAD performs well for most noise types even in low SNR but, fails for some specific noises.…”

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

“…A voice activity detector basically consists of two main processes, namely feature extraction and classification. Some of the popular features in speech processing are zero crossing rate [4], energy [5], signal-to-noise ratio, spectral flatness [6], correlation [7], etc. Instead of modeling the dynamic noise features using support vector machine (SVM) trained on noise-labeled training data [8], some recent VADs focus on the extraction of robust speech features such as the formant frequencies of eight English vowels [9].…”