Forums

Microphones position for sound source localization

Started by lapidarylee May 8, 2009
Hello,

I've recently studied some papers about sound source localization method.
As I know, 3D sound source localization can be done by Time Difference of
Arrivals (TDOA) & Generalized Cross Correlation (GCC) methods with four
microphones placed on different plane.

Its a question about microphone position, how can I place them well for
optimization ?!

If the environment is a public square outside, can I put them at corners,
spaced more than 10m apart from each other ?!
This question arised coz all papers which I surveyed keep them apart just
no more than 50cm. 
I dont know whats the effect of putting them at so different distance. 

Thanks

Ryan

lapidarylee wrote:
> Hello, > > I've recently studied some papers about sound source localization method. > As I know, 3D sound source localization can be done by Time Difference of > Arrivals (TDOA) & Generalized Cross Correlation (GCC) methods with four > microphones placed on different plane. > > Its a question about microphone position, how can I place them well for > optimization ?! > > If the environment is a public square outside, can I put them at corners, > spaced more than 10m apart from each other ?! > This question arised coz all papers which I surveyed keep them apart just > no more than 50cm. > I dont know whats the effect of putting them at so different distance.
General antenna theory applies. Beam width ~ wavelength/spacing. However if the spacing is more then half of the wavelength, then the directional pattern will have multiple beams. You will have to resolve the ambiguity. Vladimir Vassilevsky DSP and Mixed Signal Design Consultant http://www.abvolt.com
On May 8, 3:17 am, "lapidarylee" <Ryan_...@ali.com.tw> wrote:
> Hello, > > I've recently studied some papers about sound source localization method. > As I know, 3D sound source localization can be done by Time Difference of > Arrivals (TDOA) & Generalized Cross Correlation (GCC) methods with four > microphones placed on different plane. > > Its a question about microphone position, how can I place them well for > optimization ?! > > If the environment is a public square outside, can I put them at corners, > spaced more than 10m apart from each other ?! > This question arised coz all papers which I surveyed keep them apart just > no more than 50cm. > I dont know whats the effect of putting them at so different distance. > > Thanks > > Ryan
When doing TDOA with GCC via DFT by the most straightforward approach, remember that the DFT time window must span an interval big enough to hold the duration of the signal you with to locate plus the longest element to element transit time at speed of sound. Dale B. Dalrymple
On May 8, 3:17&#2013266080;am, "lapidarylee" <Ryan_...@ali.com.tw> wrote:
> Hello, > > I've recently studied some papers about sound source localization method. > As I know, 3D sound source localization can be done by Time Difference of > Arrivals (TDOA) & Generalized Cross Correlation (GCC) methods with four > microphones placed on different plane. > > Its a question about microphone position, how can I place them well for > optimization ?! > > If the environment is a public square outside, can I put them at corners, > spaced more than 10m apart from each other ?! > This question arised coz all papers which I surveyed keep them apart just > no more than 50cm. > I dont know whats the effect of putting them at so different distance. > > Thanks > > Ryan
Tricky - the more mics the better. Depends on the locus of the unknown source wrt the array. Hardy