Menu

How Can I obtain MFCC

2010-10-24
2012-09-22
  • Marlon Chalegre

    Marlon Chalegre - 2010-10-24

    i'd like to do this:

    "The speech utterance was segmented into 16ms frames, and
    Hamming window was applied for smoothing and reducing spectral leakage. A
    frame was overlapped with the next frame by 9ms. Each segmented speech frame
    is parameterized using 12 Mel-frequency subband energy values. The observed
    speech frame is transformed to the frequency domain using FFT. Then, a set of
    12 Mel scaled filter banks which has frequency span between 200Hz to 3.2kHz,
    was applied to the FFT power spectrum.(...) The vector of 12 speech power
    coefficients for each speech frame, nf , was assigned to a cluster by vector
    quantization"

    I am looking for a example about how obtain MFCCs using Sphinx4. I wanna
    obtain the vectors and pass it for a classifier (intelligence computacional
    algoritm).

    P.S. Audio processing is not my atuation area :)

    Thanks and Sorry for my bad english.

     
  • Nickolay V. Shmyrev

    See edu.cmu.sphinx.tools.feature.FeatureFileDumper class

     
  • Nickolay V. Shmyrev

    Next time, please use sphinx4 forum when you ask about sphinx4.

     
  • Marlon Chalegre

    Marlon Chalegre - 2010-10-26

    Sorry about that and thanks for the answer you help me a lot!! ;)

     

Log in to post a comment.