Initial Ideas

1. Audification? But the data is low-pass filtered. Even is we speed up by factor 10 (duration: three seconds), the fundamental frequencies are around 100 Hz. Pitch shifting could be interesting?
2. What about slowing it down? Irregularities will be audible, perhaps using markers. Temporal perception instead of spectral perception. Ideally we can distinguish the bursts.
3. Use the product of the two variables? O1 - O2 representing different hemispheres that are not necessarily in the same phase. Then: the product is potentially misleading.
4. a) mapping: one channels (o1) -> amplitude of the fundamental; map the second to the amplitude of the second harmonic. Then you hear whether it is synchronous or not. Synchrony -> rhtyhms, nonsynchrony -> octave beats. between -> timbral transients. b)Use the two channels 1st pulse generator. 2nd filter to change the tone. Can pd make it? Yes.
5. It could be interesting to have a delayed copy of the channel - to bring a sample into reference with the signal 25 samples before and thus produce auto-correlation

Experimenting

Control the pitch of a sawtooth wave by one variable & then use the second as filter channel. Reduce the pitch range to about one octave (48-60). to strengthen perception, map to both the pitch and the amplitude.

There are in general two ways: the ideas on the one side, the tool and its capacities on the other side. While the programmers work: concentrate on the side of ideas: What features of auditory display do we want to apply here? What about spectral characteristics? Further our ideas.

Idea: Comparing data set to constant harmonic with fundamental frequency?

Suggestion: instead of mapping the values - map the square or a higher power of the values.

Seek for the rhythm? Sharpen the peaks by mathematical procedures. However, from the potential on the outside you cannot compute the potentials on the inside. The problem is that you cannot find an absolute phase.

There are two principle approaches, one with a continuously changing sound and one event based: to get the rhythm rate. When you have a continuous sound, the fatigue is a much more important problem than with the event based approach.

Evaluation

We found that this (pd) toolkit is limited. It takes time to program.

Whats next?

Non-linear distortion is promising.

Working with delayed signals is promising.

The clipping of negative values to zero is promising.

Maybe abandon concept of rhythmic signal to explore some other qualities of the data: pitch change in data? More harmonics in data?