Exploiting ASR in CASA : ICSI talk, 1997may21


These are the slides I used in a talk I gave at ICSI on 1997may21. The talk described my WASPAA'97 paper, Computational Auditory Scene Analysis using speech recognizer knowledge.


  1. Exploiting Automatic Speech Recognition (ASR) knowledge in Computational Auditory Scene Analysis (CASA)
  2. Analyzing sound mixtures
  3. Integrating ASR into CASA
  4. Inverting speech representation
  5. Inverting labels to features
  6. Inverting RASTA (features->spectra)
  7. Preliminary results
  8. Preliminary results 2
  9. Conclusions

DAn Ellis <dpwe@icsi.berkeley.edu>
International Computer Science Institute, Berkeley CA