I'm working to prepare research article for my project. While preparing for it, I've gone through the topics like Gaussian mixture model and Fourier transform for voice classification problems. I've got sounds from the sources(probably the recorded voice of the different peoples).

Can I perform DFT on those audio files and pass the points from the DFT to the the Gaussian mixture to extract the features from the audio files?

Later the weights from the Gaussian mixture model will be used to compare the sounds.

  • $\begingroup$ You can transform anything. Sometimes even meaningfully! Point being: yes, you can certainly do that. It should be pretty straightforward to do and try (so, try, learn something, and if worthwhile, write something about what you've learned)! Since you don't tell us anything about the goal of your project, it's hard to tell how promising that approach is. The DFT alone isn't very often used in voice analysis; have you looked at other feature sets like MFCCs? $\endgroup$ – Marcus Müller Jan 24 '19 at 10:03
  • $\begingroup$ What are the mathematical prerequisite for understanding mathematics of MFCC feature extraction? I've gone through digital signal processing lecture(Z-transform, DFT, FFT). $\endgroup$ – sbhusal123 Jan 27 '19 at 4:48
  • $\begingroup$ It should be pretty easy to understand with your background. Wikipedia is your friend. $\endgroup$ – Marcus Müller Jan 27 '19 at 11:38

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.