Personal tools
You are here: Home Science by Ear Workshop Problem sets Ultra Wide Band Communication Session Notes

Session Notes

Notes taken during this session.

Initial Ideas

UWB Data. Data contain deterministic and noisy components. The components are distinguishable according to their mean and variance.

Deterministic components stand out and are of importance but there is considerable energy in the noise components so they might also carry valuable information

Sweet spot control: mirrro data such that an object in quest puts you in the sweet spot when present.


listening to audified data creates clear auditory objects of the UWB response but deterministic components are not salient.

It turns out that the noisy components were amplified compared to the det. comps. Correcting for this creates new objects by amplitude modulation of two oscillators. First two channels are "loss",

Is it possible to use the mean to differentiate? This would be even better if realized interactively by taking the angle into account. One would like to choose a location and listen to the response. However, data of the corresponding images are not available in the first session.

The det. comp. would then possibly create salient aud. objects.

Christopher has previously tried to use the shapes of the responses for sonification because they are rather characteristics.

Shape features may form a side features

The preprared angle-delay graphics of the "energy" suggests event-based sonification. The delay might be kept as a temporal axes for the sound.

The shapes of the deterministic components may be used to create a sharper focussed or more diffuse sound.

Angle could be a pitch, Intensity could be loudness, shape would produce different resonances, something tight goes "boig" something wider would go "boiiiiiinnnnnnggggg"

However shapes are quite similar in spectrum, so their distinction would have to be refined. Information could be put into the attack of the sound.

Afternoon Session

Sandra: Compare the two raw signals in the amplitude-modulated approach, yields decaying power after the initial peak but does not reveal more structure due to the linear mapping to amplitude.

Try fourth power of signal to nonlinearly distort it and make events more salient.

Thomas: Try to use the shape to filter noise into colored noise. Yields different noises but not easy to recognize as characteristic objects.

Try a mapping of the angle-delay-power-shape data:

  • time->time
  • angle-> panning
  • magnitude-> intensity
  • shape-> sinewave modulation

Time could go into the frequency of a rhythm and angle into a pitch Magnitude can go into pitch: first range is too extreme.

We would like to have the events triggered by the delay time but that requires previous ordering.

Test: follow the time course, trigger det. components at the time they were recorded, their power is mapped to pitch, their shape to brightness (or color).

Due to the use of only a single oscillator, events that are very close to each other are cancelled by the successors. This is solved by giving to each event a minimum duration.


Powered by Plone CMS, the Open Source Content Management System

This site conforms to the following standards: