Pink Trombone (Experiments with Google)

Interactive simulation of the human vocal tract.

Moving several points of interaction around controls the sound which comes out of the synthesizer. The parameters are modeled after the parts that are believed humans are able to consciously control. This affects subparameters of the model which synthesize human-like sounds.

Quinn: this is of interest because we should be able to teach a reinforcement learning model to operate the virtual larynx.