Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

In no way at all, especially statistical techniques for speech enhancement using Emax or Bayesian given Gamma prior.

You can also toys in some more advanced stuff like ICA deconvolution.

This on top of some nice improvements in plain DSP techniques. (E.g. modified gammatone transform or variants of Stockwell transform)

All in low latency.

The main problem is ruining this on a tiny jellybean power efficient micro. Deep learning on a Cortex M0 with decent quality and less than 5ms latency? Good luck.



I read the suggestion as "applying deep learning to optimize the parameters of the process", not running it on the device itself.


For optimization of parameters you can go full ham and apply even genetic algorithms.

The problem is actually finding a "legibility" and "quality" metric. Codec ones like PEQ are not good enough.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: