I am doing a machine learning project and I'm interested in the different pre-processing techniques which can be applied to audio data. In particular, I am interested in comparing a human auditory model (basically pre-process my data into features which mimic how humans perceive sound) and a non-human auditory model. I know for example that the Mel Spectrogram is a human-auditory model. Or the multi resolution cochleagram. But what are the ones that are not human auditory? I suppose an example is the short-time Fourier transform? Or the PCM data?

What would be other examples?

  • $\begingroup$ JTFS is mammalian. Code to be released. Since CWT-based models are good human auditory models, and it is significantly different from STFT, I suppose you could call STFT "non-human". $\endgroup$ Commented Jun 20, 2022 at 19:37
  • $\begingroup$ If you say "non-human" do you mean other live forms or mathematical (non-biological) algorithms? $\endgroup$
    – Hilmar
    Commented Jun 20, 2022 at 21:28
  • $\begingroup$ @Hilmar yes, exactly, mathematical (non-biological). Do you know others of the types? $\endgroup$
    – Slim Shady
    Commented Jun 21, 2022 at 8:35


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.