How to extract features from speech signals using MFCC?

59 views (last 30 days)
NIROSHA M
NIROSHA M on 6 Mar 2015
Answered: Ashwini P on 1 Jun 2020
I am using isolated words as my input speech signals.I have a speech signal of length 1.8193sec that contains 14554 samples. I have done pre-emphasizing of the signal.I have obtained 91 frames with 160 samples per frame. Now I have to do frame overlapping and windowing.I am confused here . I don't know which window function is appropriate for my signal.Please help me to do this.
  1 Comment
Md Sahidullah
Md Sahidullah on 4 Jun 2015
Hi, For step-by-step calculation for MFCC features, you can check this paper: Md. Sahidullah, Goutam Saha, Design, analysis and experimental evaluation of block based transformation in MFCC computation for speaker recognition, Speech Communication, Volume 54, Issue 4, May 2012, Pages 543-565, ISSN 0167-6393, http://dx.doi.org/10.1016/j.specom.2011.11.004. http://www.sciencedirect.com/science/article/pii/S0167639311001622
For the selection of window for MFCC, check this paper: Sahidullah, M.; Saha, G., "A Novel Windowing Technique for Efficient Computation of MFCC for Speaker Recognition," Signal Processing Letters, IEEE , vol.20, no.2, pp.149,152, Feb. 2013 doi: 10.1109/LSP.2012.2235067
Hope this helps. Thanks Sahid

Sign in to comment.

Answers (3)

Idris Al-ismaili
Idris Al-ismaili on 17 Jul 2018
pleas, i want code for detect baby crying i am working on intelligent cradle project

Brian Hemmat
Brian Hemmat on 12 Feb 2020
Starting from R2018a, you can use the mfcc function in Audio Toolbox to extract mel-frequency cepstral coefficients.

Ashwini P
Ashwini P on 1 Jun 2020
i want an Matlab code to extract features fro a speech which is just a word lasting for maximum of 0.2secs

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!