Extract feature from audio file using DWT

I have an assignment of extracting features from audio file using Discrete Wavelet transform and C++. I believe the purpose of the homeworks were just for demo, so here is my plan for it:

  1. Audio file is .wav file. I will read .wav file as an array of signals.
  2. After getting signals array, i apply Discrete Wavelet transform using Haar
    wavelet to get the coefficients array. Then i use this array to calculate
    features.

Since my knowledge is limited, i have some questions:

  1. How to read .wav file by C++ properly?
  2. I know if Haar wavelet decomposition level is $k$ then signal’s length
    should be divided by $2^k$. So what should we do if signal’s length is
    not divided by $2^k$?.
  3. Which features can we get from the coefficients array of DWT method
    above (if the method were right)?
  4. I read some documents which said if length of signals array was loo
    long then we should apply DWT on overlapping segments. Does it
    change how we calculate the features?

I know that i should search for the anwser first. But since i don’t have much time for researching, i really need your help.

Leave a Reply

Your email address will not be published. Required fields are marked *