camera tracking of a single sound source
camera tracking of a single sound source
(OP)
im trying to bulid a system which steers a camera based on the speaker location (sound direction of arrival actually) obtained by a microphone array using the Steered Response Power - Phase Transform (SRP-PHAT) algorithm, i have a few problems :
A. i dont know how to choose the right microphone array configuration (linear, circular...etc) i cant see a difference between them since i just need to calculate the generalized cross correlation between their signals.
B. im confused between choosing Arduino or Raspberry Pi to control motors used to steer the camera, which one is more suitable to control two servo motors ???
A. i dont know how to choose the right microphone array configuration (linear, circular...etc) i cant see a difference between them since i just need to calculate the generalized cross correlation between their signals.
B. im confused between choosing Arduino or Raspberry Pi to control motors used to steer the camera, which one is more suitable to control two servo motors ???





RE: camera tracking of a single sound source
A. Why do you need more than two?
B. I would be more concerned about throughput.
TTFN

FAQ731-376: Eng-Tips.com Forum Policies
Need help writing a question or understanding a reply? forum1529: Translation Assistance for Engineers
RE: camera tracking of a single sound source
A.i need more than two in order to estimate the location of a source in 3d space.
B.in terms of processing Raspberry Pi is better but i dont know if its capable of controlling two motors?
RE: camera tracking of a single sound source
TTFN

FAQ731-376: Eng-Tips.com Forum Policies
Need help writing a question or understanding a reply? forum1529: Translation Assistance for Engineers
RE: camera tracking of a single sound source
Is it a predefined sound at a given frequency? If so, the usual default sensor spacing (in general) is one-quarter lambda.
If not, then the problem is infinitely more complicated. Because it's undefined, covers several orders of magnitude (20-20,000 Hz), allows multiple solutions, etc.
Pulse or impulse waveforms (gun shots) are pretty easy, obviously. Spacing based on your timing resolution.
If it's 3D, you'd need three baselines, minimum of four sensors (mics). Origin, X, Y, and Z.
This is all very standard.
RE: camera tracking of a single sound source
You need 4 microphones worst case, 3 if you aren't worried about the back plane. Arrange them as you like and do a bit of maths. Obviously the most elegant algorithmically is as a tetrahedron.
More mics don't really help much unless you are concerned with redundancy.
Is this in a reverberant room? that could be quite fun.
Cheers
Greg Locock
New here? Try reading these, they might help FAQ731-376: Eng-Tips.com Forum Policies http://eng-tips.com/market.cfm?
RE: camera tracking of a single sound source
I vote for the Arduino. Of course the Arduino doesn't have the CPU power to calculate your SRP power.
You may need both !!
Capture the Audio with Arduino and pass it on to the RPi for the serious number crunching and it returns
desired motor locations for the camera is how I would do it with what you have specified.