Here is a small outline of the steps involved in doing this Perl.
first convert your mp3 file into a .wav file with the lame encoder (-d option).
Using the Win32::Sound module load this .wav file and read the samples.
Take a FFT to convert your time signal into frequency domain
Optionally, you can use a comb filter (it involves convolving a series of sample signals to determine which signal yields the highest energy) as described in the above paper. Personally, I skipped it
Cheers!
san.
print length "The answer to life, universe & everything!"