Your question is not clear. I'm guessing that you have a file of audio samples like a .wav file of PCM data.
If your data contains most of it's energy at one frequency eg sine or sqaure wave you can process the samples to remove any DC offset then count the number of samples between zero crossings.
If you have data such as a recording of voice and music then it contains energy at lots of frequencys.
A single number does not accuratly describe it. You have to run it through a fourier transform work out the frequency content.