vad

vad copied to clipboard

#+AUTHOR: Maarten Versteegh

• vad This is a straight-forward re-implementation of Bowon Lee's Voice Activity Detector. The relevant paper to cite is: Bowon Lee and Mark Hasegawa-Johnson, "Minimum Mean-squared Error A Posteriori Estimation of High Variance Vehicular Noise". Please cite that paper when using this code for research purposes.

** Usage : >>> from vad import VAD : >>> detector = VAD(fs=16000) : >>> speech = detector.detect_speech(sig, threshold=0.5)

** Requirements

• python>=2.7 / python>=3.0
• numpy>=1.8.0
• scipy>=0.14.0