next up previous
Next: Experiment 2: Multiple Model Up: Experiment 1: Single Model Previous: Results

Discussion

These results show that an estimator can be trained on a non-exhaustive data set and can extract physically meaningful sounds from trained data as well as novel sound data. We now go on to generalize this technique to estimating for multiple models simultaneously.



Michael Casey
Fri Mar 22 15:49:22 EST 1996