If I've got a bunch of short .wav files, 16 bit linear pcm, and an incoming stream of audio that's been run through a voice codec (g.722, g.711, or others) and then converted back to pcm, what's the right algorithmic tool to detect when one of the wav samples is present in the audio?

The context for this is automated testing of a telephone IVR system, where I want to know that certain sounds have been successfully played.

Again, this is for detecting a known list of sounds in audio that has been distorted by voice codecs. This seems like it should be an easier problem than voice recognition, but I haven't had much success in past attempts.

  • 1
    $\begingroup$ Have you heard of cross-correlation? $\endgroup$
    – Jdip
    Commented Aug 16, 2023 at 17:29
  • 1
    $\begingroup$ Here are two closely related answers. $\endgroup$
    – Gillespie
    Commented Aug 16, 2023 at 17:51


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.