As a part of our project we should classify sound samples (stored as .wav files). All sample is the same, just pure speech (like a Skype test call).

The process is the following:

  • a reference wav, this is the "high quality" sample

  • comparing approx. 1000 wav files

  • calculating divergence from the reference wav one by one

Is there any Linux tool for that?

  • Do you need only to compare the files, or you want to listen to the differences? – Luciano Andress Martini Jan 20 '17 at 10:43
  • Yes, it's about "hearing" – sdkféldf werer Jan 20 '17 at 12:39

What I believe you are trying to measure (by stating divergence) is the PESQ, Perceptual Evaluation of Speech Quality, of each file. This is a standarized form ITU-T recommendation P.862 (02/01) http://en.wikipedia.org/wiki/PESQ.

You have different projects implementing what you are searching for. For example https://github.com/imankulov/speex-quality-evaluation

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.