Optimizing Time-Frequency Masks for Robust Missing Data Speaker Recognition
A PhD Seminar by Daniel Pullella, Signals and Systems Engineering Research Group (SSERG) School of EE&C Engineering, UWA
| What | Seminar |
|---|---|
| When |
2008-10-09 13:00
2008-10-09 14:00
2008-10-09 from 13:00 to 14:00 |
| Where | Billings Room (3rd Floor) EECE |
| Contact Name | Dr. Roberto Togneri |
| Contact Email | roberto@ee.uwa.edu.au |
| Contact Phone | 6488 2535 |
| Add event to calendar |
|
Despite their effectiveness for robust speaker recognition, missing data techniques are vulnerable to errors in the classification of the input speech signal's time-frequency points. A direct method for the removal of these mask errors is through the top-down optimization of the estimated mask, however this requires a measure to evaluate the mask quality without a priori noise knowledge. This research proposes the normalized likelihood confidence as such a criterion. In this approach ......