rVADfast 0.0.1

rVADfast - a fast and robust unsupervised VAD

rVADfast

The Python library for an unsupervised, fast method for robust voice activity detection (rVAD), as presented in the paper rVAD: An Unsupervised Segment-Based Robust Voice Activity Detection Method, Computer Speech and Language, 2020. More info on the rVAD GitHub page.

The rVAD method consists of two passes of denoising followed by a VAD stage. It has been applied as a preprocessor for a wide range of applications, such as speech recognition, speaker identification, language identification, age and gender identification, self-supervised learning, human-robot interaction, audio archive segmentation, and so on as in Google Scholar.

The method is unsupervised to make it applicable to a broad range of acoustic environments, and it is optimized considering both noisy and clean conditions.

The rVAD (out of the box) ranks the 4th place (out of 27 supervised/unsupervised systems) in a Fearless Steps Speech Activity Detection Challenge.

The rVAD paper is among the most cited articles from Computer Speech and Language published since 2018 (the 6th place), in 2023.

Usage

The rVADfast method is available as a python package installable via: pip install rVADfast. After installation, you can import the rVADfast VAD class as from rVADfast import rVADfast from which you can instantiate a VAD instance, e.g. as vad = rVADfast().
The package also contains functionality to process folders of audio files, to generate VAD labels or to trim non-speeh segments from audio files. This is done by importing the rVADfast.process module which has two methods for processing audio files, namely process.rVADfast_single_process and process.rVADfast_multi_process, with the latter utilizing multiple CPUs for processing. Additionally, a processing script can be called from commandline-tools by executing ```rVADfast_process

In /notebooks a concrete example on how to use the rVADfast package is found.

References

1. Z.-H. Tan, A.k. Sarkara and N. Dehak, "rVAD: an unsupervised segment-based robust voice activity detection method," Computer Speech and Language, vol. 59, pp. 1-21, 2020.
2. Z.-H. Tan and B. Lindberg, "Low-complexity variable frame rate analysis for speech recognition and voice activity detection,” IEEE Journal of Selected Topics in Signal Processing, vol. 4, no. 5, pp. 798-807, 2010.