As a part of my thesis work -in which I am using both acoustic and visual information- I am looking forward to implement a real time sound source localization sub-system. From the geometric principles, I know that at least 4 non-coplanar microphones are needed for 3D localization.

However, as I understand, using only a laptop to interface 4 microphones is impossible. I read that one sound card can only support 2. I also search through 4 microphone usb interfaces and found out quiet expensive solutions. I have also encountered to PS Eye, I'm not sure its sensory information is accessible through MATLAB or CLAM/C++ though.

As I only need an approximate information, I then decided to use only the azimuth angle information (planar confidence) of the source. For that, I know 2 microphones are enough. But I read about another theoretical problems for this case, like the need of calibration, harder sound seperation etc.

I have to ask is there a cheap interfacing solution, using 3 or 4 USB microphones? If not, would it be feasible to build up something from scratch and use amplified electret mics? I am really confused about hardware selection. Neither my advisor, nor the academic papers have too much practical information.

Thanks in advance.


Just get a multi-channel recording interface and some regular analog microphones. For example http://www.sweetwater.com/store/detail/Scarlett18i20/ has plenty of channels and microphone pre-amps.

  • $\begingroup$ Using a Kinect to access the built-in mic array would be a cheaper solution than this.. $\endgroup$
    – baci
    Jan 24 '14 at 7:51
  • $\begingroup$ +1 A multi-channel interface is better than a bunch of separate devices if you don't want to worry about mucking about synchronising the signals. $\endgroup$ Feb 24 '14 at 11:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.