As is said in the latest LIGO press realease, the detected gravitational waves vary with a frequency that is audible for us. The audio one can obtain from the detected data is actually plaied during the press realese: see https://www.youtube.com/watch?v=aEPIwEJmZyE&feature=youtu.be&t=2754 at 45:54.
I tried to reproduce this same sound, starting from the public data of the event: https://losc.ligo.org/events/GW150914/
What I do is to get the raw data from https://losc.ligo.org/s/events/GW150914/P150914/fig1-waveform-H.txt. This dataset contains only $3400$ rows, so I add new data by adding linearly interpolated data.
A sample of the python script (needs python3) is here: https://gist.github.com/anonymous/2f5f7d1824c11d2698b1
Unluckily I get something ways more noisy and "flat" than the one presented in the press release.
Do someone know which technique they used? maybe they have a sample with more datapoints in time?