This demo seems to do just that: http://acko.net/files/audiosynth/index.html http://acko.net/files/audiosynth/index.html Edit: It's not exactly as "dynamic" a

This demo seems to do just that: http://acko.net/files/audiosynth/index.html http://acko.net/files/audiosynth/index.html Edit: It's not exactly as "dynamic" as I thought. The different samples are generated once and then played in a loop. So the audio data is not generated in real time, but it is generated.