Specifying DSP for voice processing

[MEturnedEE]

Hi there, we are looking to use a DSP for voice enhancement. It will be on board a wireless microphone device which will pick-up a user's voice and attenuate considerably environmental noise. There are lots of options out there - from hardware component manufacturers which provide off-the-shelf solutions to more sophisticated software providers with proprietary algorithms. We've had numerous discussions with solution providers and all are convinced they have what we need. We've started playing around and testing with a few but hoping to approach this in a more systematic way though. What would be a suggested starting point to define the performance requirements (from MIPs, RAM, bits, clock speed, etc) for our use case?

 

[itsmoked]

Wow. You've got it all backwards now.

That is a topic you could spend years on trying to get somewhere optimal. I'd absolutely look for off-the-shelf solutions already completely done. If you've found vendors saying they have solutions - try their solutions. Try them all. Pick the one with the best price performance and support to get your mission completed. Once you have one you like THEN you will understand what platform you'll be needing to support it.



Keith Cress
kcress -

http://www.flaminsystems.com

 

[MacGyverS2000]

The algorithm to pick out one user and attenuate all others is not trivial in anything but a highly controlled environment... do as Keith suggests and try off-the-shelf solutions before spending half of your career trying to reinvent the wheel.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[MEturnedEE]

Thanks for the input!

 

[sreid]

Take a look at Microchip.

http://www.microchip.com/design-centers/audio-and-speech