dmngu9/voice-activity-detection

Efficient voice activity detection algorithms using long-term speech information in C++

This repo is for voice activity detection algorithm (VAD). This code is based on this paper:

Ramırez, Javier, José C. Segura, Carmen Benıtez, Angel De La Torre, and Antonio Rubio. "Efficient voice activity detection algorithms using long-term speech information."

And the Matlab code version (Thanks for the isrish making such a beautiful program):

https://github.com/isrish/VAD-LTSD

This cpp file take input audio data from txt file and write results to 'example*.txt' file. At the end, I use gnuplot to plot the result and it works exactly the same as the Matlab Code from isrish.

Compile and run

You need FFTW and Eigen3 libraries.

$ mkdir build & cd build
$ cmake .. & make
$ ../bin/./vad

Visualizing the result

$ gnuplot
$ > set multiplot
$ > plot "sound.txt" with 1:2 lines
$ > filename(n) = sprintf("example_%d.txt", n)
$ > plot for [i=1:10] filename(i) using 1:2 with lines

voice-activity-detection's Issues

How to install ?

I found this is a nice work!
But can you explain how to install and set Eigen3( I can't set it correctly, which has killed me hours...) correctly to work? It's just so hard for python-only to compile this , and I just really want to see the result, thanks.

The VAD.c is loss

After I download the the C code , I can only find the vad.h. The vad.c is dismiss.

What wrong with it?

License?

Hi!

What’s the license this code has been released under?

Thanks. :)

what's the sample rate of the sound.txt ?

Dear,

What's the sr of the sound.txt ?
Thx

Sample rate other than 8 kHz

Does the algorithm handle different sample rates and or frame lengths? We have 44.1 kHz sampled data and a frame length of 20ms.

The meaning of NFFT"2" ?

Hello,

What does the "2" of the variable "NFFT2" used in your VAD.h mean ?

Thanks a lot!

David