On 23/05/14 09:33 PM, Bruce Perens wrote:
> We had a chat with Jean-Marc about this long ago, but I can't find the
> message. He claims a 3 dB improvement per additional microphone (which
> is maybe why we see between 4 and 8 in modern systems). He has a number
> of papers on this issue, deriving from his work on robotic audition. See
> http://people.xiph.org/~jm/papers/IROS04_711.pdf and there is a papers
> list at http://scholar.google.com/citations?user=l52ospIAAAAJ&hl=en
Not quite. The theoretical improvement from a delay-and-sum beamformer
(*) is 3 dB every time you *double* the number of microphones, in other
words, 10*log10(nb_mics).
Cheers,
Jean-Marc
(*) This is assuming the noise is diffuse and uncorrelated between mics.
Smarter beamformers can do better for directional sources, but you can
also end up with worse if the mics are too close.
------------------------------------------------------------------------------
"Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE
Instantly run your Selenium tests across 300+ browser/OS combos.
Get unparalleled scalability from the best Selenium testing platform available
Simple to use. Nothing to install. Get started now for free."
http://p.sf.net/sfu/SauceLabs
_______________________________________________
Freetel-codec2 mailing list
Freetel-codec2@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freetel-codec2
