waveform resynthesis

BETA BETA BETA !!! — The method is still in development, but I must try to use our “waveform resynthesis” methods to deconstruct and reconstruct the LIGO data, the first measurement of gravitational waves emitted from two black holes that merged. Here is Brian Greene’s nice explanation, https://www.youtube.com/watch?v=s06_jRK939I , and the great NSF news conference: https://www.youtube.com/watch?v=aEPIwEJmZyE with a lot of information on the measurement, the phenomenon and its historical importance. Here’s the data from the Hanford, WA site: 

Here is the “direct sonification” of the data– it is a thud that is really a chirp, in a fraction of a second.  The data is here: https://losc.ligo.org/events/GW150914/

So, because it passes so quickly, we want to instead be able to listen to the phases, the individual events in the waveform. In this ”waveform resynthesis” method, we filter the signal many times with narrow band pass filters, then extract from each filtered signal its amplitude envelope, and then use each envelope to modulate the amplitude of a sine wave oscillator (a multiplication of the envelope with the oscillator. Then all of these oscillators (in the “oscillator bank”) are summed up to create a new signal that has characteristics of the initial signal, but many many more cycles so that individual phases are now sound events. We can hear individual phases by filling them up with summed waves of higher frequency, that we can stretch out as long as we want.

Here are the filters:









And here is the set of filtered signals:


Here are the frequencies at which we filter on the x-axis, and frequencies of the oscillators that will get modified on the y-axis. These are arbitrary aesthetic choices.








And the summed oscillators in the new signal:


And the new sound, 12 seconds long, with the Hanford, WA data in the left channel and Livingston, LA in the right channel:

The low beating pulses are, or should be, I think, the individual phases of the gravity waves. The higher frequency noise sloshes around incoherently between the stations, but the low frequency beats are coherent, increasing in frequency and amplitude. There is a lot of spurious information introduced, but in a sense that is what we are trying to do– so there is a lot of work remaining to control what information is introduced.



Leave a Reply