Forums

Position estimation to sound source

Started by analyser August 7, 2006
Hi, I am doing this final year project in estimating the position of sound
source. I have constructed two detection sets with each of four
microphones. Each of these detection sets will yield a direction. Knowing
that distance apart, we can localize the sound position in space. 

The problem now is I have to consider the direction of each detection set
first. How do I calculate out the azimuth and inclination angles using the
time delay?



analyser wrote:
> Hi, I am doing this final year project in estimating the position of sound > source. I have constructed two detection sets with each of four > microphones. Each of these detection sets will yield a direction. Knowing > that distance apart, we can localize the sound position in space. > > The problem now is I have to consider the direction of each detection set > first. How do I calculate out the azimuth and inclination angles using the > time delay?
A similar problem is faced in GPS. Google for GPS estimation and you may find the answer. Baaaaa
You might want to look into split-beam cross-correlation.  Try google.

In article <e7adnasQTY4fQ0rZnZ2dnUVZ_oWdnZ2d@giganews.com>, "analyser" 
<cookiemon_29@yahoo.com.sg> wrote:
>Hi, I am doing this final year project in estimating the position of sound >source. I have constructed two detection sets with each of four >microphones. Each of these detection sets will yield a direction. Knowing >that distance apart, we can localize the sound position in space. > >The problem now is I have to consider the direction of each detection set >first. How do I calculate out the azimuth and inclination angles using the >time delay? > > >
analyser wrote:

   ...
> I have constructed two detection sets with each of four > microphones. ...
I don't think you mean that, and from the replies you got, neither do others. You seem to have meant "I have constructed two detection sets, each with four microphones." BTW, at the price of a little more computation to get the beam directions, three microphones per set are enough. Jerry -- Engineering is the art of making what you want from things you can get. &#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;
sheepshaggerx@yahoo.co.uk wrote:
> analyser wrote: > > Hi, I am doing this final year project in estimating the position of sound > > source. I have constructed two detection sets with each of four > > microphones. Each of these detection sets will yield a direction. Knowing > > that distance apart, we can localize the sound position in space. > > > > The problem now is I have to consider the direction of each detection set > > first. How do I calculate out the azimuth and inclination angles using the > > time delay? > > A similar problem is faced in GPS. Google for GPS estimation and you > may find the answer. > > Baaaaa
This is NOT the way to proceed. There may be "similar" problems in GPS (I am not sure about that though, but leave that for now), and whatever solutions are used there can not be used for sound source localization. In GPS, the source (the satellites) send information which contents and encoding is known to the reciever. In source localization, one usually have to estimate both the presence and direction of a source. As for the OP's task, there is a large body of literature on planar arrays. It's not what I am most familiar with, but ther is a book by Ziomek (don't remember the title) that ought to provide a good starting point. Rune
"analyser" <cookiemon_29@yahoo.com.sg> wrote in message 
news:e7adnasQTY4fQ0rZnZ2dnUVZ_oWdnZ2d@giganews.com...
> Hi, I am doing this final year project in estimating the position of sound > source. I have constructed two detection sets with each of four > microphones. Each of these detection sets will yield a direction. Knowing > that distance apart, we can localize the sound position in space. > > The problem now is I have to consider the direction of each detection set > first. How do I calculate out the azimuth and inclination angles using the > time delay? >
With four microphones, you should be able to calculate the direction (bearing and elevation) to a sound source relative to some assumed zero direction, say a vector from the center of the array out through microphone #1. By itself, this does not tell you anything about the orientation of the microphone array. If you know this orientation, then you can get a direction to the sound source in the real world. Then, given two oriented arrays at known positions, you should be able to determine the position of the sound source. There are several ways to determine the array orientation: 1. You can manually orient the array when you set it up ( set it flat on the ground oriented to the north for instance) 2. You can attach some kind of computer readable compass and tilt sensor to each array. 3. You can start with the sound source in a known location and back out the array orientations. BTW, based on having done this sort of thing for a living, you are probably better off working in the frequency domain and thinking in terms of phase differences rather than than time delays. If you are going to test outside, don't forget some big wind screens Best wishes, --Phil Martel