I have been trying to find an article where it is described how real-time (or close to real-time) signal processing from a microphone is able to detect that an audio source is playback of a recording rather than natural source thanks to (for example) missing frequencies in lossy encoding. Ideally it would be something about voice authentication tampering / voice assistant trigger words filtering. I am sure there is lots on the subject, however I am quite new to the field and can't seem to be able to come up with the right keywords for my googling.


  • $\begingroup$ Frank Sinatra is dead, so if a speaker id algorithm identifies Frank Sinatra, its recorded. My point is that it iis doubtful that a single solution exists for every case. You could search on LP or tape artifacts and perhaps solve part of the problem, but a recording from a well mastered CD would require a different approach $\endgroup$ – user28715 Feb 3 '18 at 11:35
  • $\begingroup$ There is a whole lot of research going on in the field of speaker identity verification which tackles a problem of someone playing back the recording of the speech. I will add an answer later on with few approaches as I am away from home right now. $\endgroup$ – jojek Feb 3 '18 at 13:09

Your Answer

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

Browse other questions tagged or ask your own question.