novnain

Objective: Sample sounds at one end of the tunnel and throw it into the fray of sounds going through the pd-cluster.

For a short term performance with only a few discrete sounds, organizing the sounds wouldn't be very helpful. But if this installation was left up for a long period of time, I think it could be very interesting to try to parameterize all of the sampled sounds into a database, then as part of the composition algorithm, load stored samples based on various parameters.

Parameters

Upon recording a sample, the parameters would be determined by the sampler program. There would be basic parameters, like length and loudness of the sample, but I think the more useful parameters would need to come out of some audio analysis. For example, whether the sample had lots of high or low frequency content, etc.

• length of sample
• peak loudness
• average loudness
• spectral content
• rhythm

Use of Parameters

To use the parameters to compose algorithmically, a Markov distribution could be used to choose the sounds. This would provide a reasonable amount of structure while also providing some random surprises.