11-24-2014 10:39 PM
Hi,
Im doing project in sound source localization. I have started
with finding the time delay between the 2 microphones using cross correlation.
I have done by giving a simulated signal first. The problem which im facing is
, if i give a floating point value like 0.05 as a delay im not able to get the
output delay the same. But it works well for real numbers. can u plz help.
11-25-2014 03:31 AM - edited 11-25-2014 03:37 AM
You wire a value of 0.05 to an integer and expect that to result in what? 50? 5000?
Obviously, you have to dig a little deeper in frequency basics like phaseshifting, time vs. samples. vs. frequency and computational constraints (data types like integer vs. floating point).
If you are going to hardware later on:
You should use dynamic signal acquisition devices (DSA) for audio acquisition and you require high accuracy of synchronization. Did you already thought about this?
Norbert
EDIT: Please leave text in default color. Your front panel is close to useless as labels are not readable (bad contrast). Also updated link for synchronization to a better white paper.
11-25-2014 04:13 AM
Thanks for the reply . I wire a 0.05 and i should get 0.05 as output.. Im not able to analyse what mistake i have done.
My next question was the same. How to acquire 2 signals simultaneously. tHank u for the pre-answer. I have a PXI with 1 GRAS microphone and one USB microphne.. Is it possible to acquire signal at the same time?
11-25-2014 04:46 AM
You add SAMPLES as offset (phase shift) to your signals. As there is no fractional number of SAMPLES, you see no shift when adding 0.05.
You have to compute the number of SAMPLES your TIME 0.05s represent. This is depending on the SAMPLING SPEED (aka sample frequency) which you configured for simulate signal in your VI.
I doubt that you can synchronize your microphones as you don't know the delays of the sound devices and the driver for those. This can be several us up to easily more than 100ms. Taking speed of sound into account, there is an inherent (and unknown) inaccuracy of easily more than 30 m in your measurement. Depending on the desired accuracy, you require highly synchronized data aquisition including same wire lenght!
I think you can find this reference architecture very intersting!
Norbert
11-27-2014 03:36 AM
This white paper is really interesting.. thank u so much.