Objective

To create a (stereo) synthesized pad similar to a flute -- yet strange -- using granular synthesis.

Strategy

Again I'm using diskgrain for the synthesis, this time with a new idea: to create a stereo pad by slight variation of the grain size of the right and left channels. This should smooth out the 'bumps' and create an interesting stereo texture.

Materials

Again, the flute sample is from University of Iowa's Music Instrument Samples site. I cut out an F4, normalized it, then amplified to peak full scale.

Result

The first major thing I learned was probably a basic principle of Csound all along: when performing experiments, set up an instrument in the orchestra file, and vary the instrument using parameters in the score file. Makes sense. My original strategy was to create individual instruments with variations; now I see this will work better for comparison of different results and avoid unnecessary cut-and-paste coding.

I tried four different variations of the sound. All of the sounds are slightly time-shifted relative to each other (slightly different start times and durations). This keeps the exact same portion of the sound file from being read in both channels over a long period, and hopefully will smooth the sound. Other than that, the variations were:

1. Two channels, each made of 40 ms samples.
2. Two channels, each made of 50 ms samples.
3. One channel mostly 40 ms, the other channel mostly 50s
4. 40 ms and 50 ms channels, with the 'variance' the grain size on each channel altered.

Conclusion

I believe the thing I've learned most through this experiment is how to streamline my experiments, making them easier to read and easier to change and study. The flute sound is interesting, but there is still a bit of cyclic 'thrumm' that distracts from the main sound. With experience, I should be able to adapt this technique and improve on it in later versions.