# The Grey Labyrinth is a collection of puzzles, riddles, mind games, paradoxes and other intellectually challenging diversions. Related topics: puzzle games, logic puzzles, lateral thinking puzzles, philosophy, mind benders, brain teasers, word problems, conundrums, 3d puzzles, spatial reasoning, intelligence tests, mathematical diversions, paradoxes, physics problems, reasoning, math, science.

Message body

 Emoticons View more Emoticons
 [quote="extropalopakettle"]I was listening to an awesome thunderstorm today, and noticed in particular how the sound of thunder has this sort of two-dimensional texture, a single long rumble of it emanating different timbres and frequencies from different parts of the sky. Now, I have a rough idea how human stereoscopic hearing works: The brain detects the latency between when a sound reaches one ear versus the other, and from this can judge it's position, at least along a one-dimensional left-right axis (we can usually get a good sense of up versus down too by tilting our heads, something we do unconsciously). We manage to do this for many sounds happening at once, locating them all in space. Theoretically, with three microphones spaced apart, recording the same sound, one should be able to process it and calculate a relative direction from which the sound emanated. I thought it would be cool to do this, and create visual representations of the frequencies emanating from different directions. This could be done with the sound of thunder, or a tree full of birds, or any setting in which many sounds are coming from different directions. The result should be a movie, a changing two dimensional field of color, with color mapped to frequency (there would be a loss of information, like timbre). I'd imagine what I need is an algorithm which could take a sound file and convert it to data about the dominant frequencies over time. Anyone know what algorithms do this? Is Fourier transform what I want? (I'm not looking for code, just the basic math / algorithms)[/quote]
Options
HTML is OFF
BBCode is ON
Smilies are ON
 Disable BBCode in this post Disable Smilies in this post

 All times are GMT
 Jump to: Select a forum Puzzles and Games----------------Grey Labyrinth PuzzlesVisitor Submitted PuzzlesVisitor GamesMafia Games Miscellaneous----------------Off-TopicVisitor Submitted NewsScience, Art, and CulturePoll Tournaments Administration----------------Grey Labyrinth NewsFeature Requests / Site Problems
 Topic review
Author Message
cb*
Posted: Mon Mar 19, 2012 5:54 am    Post subject: 1

Courk wrote:
Wouldn't you need 4 microphones? 3 points merely define a plane.
You're right about the number, but not about the reason. To simplify things, in a plane, 2 points define a line, but that seems quite irrelevant to me, considering that having 2 mics would actually mean getting a Hyperbola. And as i just found out, that article actually says: "This definition (locus of points where the absolute value of the difference of the distances to the two foci is a constant) accounts for many of the hyperbola's applications, such as trilateration; this is the problem of determining position from the difference in arrival times of synchronized signals, as in GPS."

extro, Multilateration seems to be just what you want. It also suggests that "enhanced accuracy can be obtained with more than four receivers", although this might increase the complexity of analysis. As for sync'ing, I intuitively wouldn't expect major issues unless there was a lot of environment noise around.
Jack_Ian
Posted: Tue Jul 05, 2011 2:28 pm    Post subject: 0

Multiple computers. Don't worry about synchronising. Just make some noise from a known location after the computers are set up and use this noise to calibrate the subsequent data.
extro...*
Posted: Mon Jul 04, 2011 4:00 pm    Post subject: -1

Courk wrote:
Wouldn't you need 4 microphones? 3 points merely define a plane.

My brain is hurting thinking about it. I need a good way to visualize. For simplicity, say I'm just trying to locate the source of a single 'click' sound. Essentially, for any pair of mics, I know the difference in its distance from each of the two mics (but not distance itself).

As Zag points out, this is like GPS. The satellites have synchronized clocks and broadcast time signals. In my case, instead of matching up time signals and the differences in when they were received, I'd be matching sounds with the same characteristics (hence the need for something like a Fourier transform or spectral analysis of some sort), and calculating position (or at least direction) from the differences in their arrival time.

GPS requires 4 satellites to work, so that should certainly be sufficient. As I initially described it, I was only interested in creating a two dimensional projection of sound sources from the 3 dimensional world, but heck, why not make it 3D? The hard work for me would be just analyzing any single pair of sound data, picking out components and matching them across tracks.

And then on the hardware side, would I use a single computer with multiple sound cards (how possible is that?), or multiple synchronized computers? I need fairly precise alignment of 3 or 4 audio tracks.
Courk
Posted: Mon Jul 04, 2011 7:01 am    Post subject: -2

I'd think sooner than a year -- after a year you're already over half a mile off the mark.
Zag
Posted: Mon Jul 04, 2011 4:44 am    Post subject: -3

This is essentially how GPS works, except that it is radio waves, not sound waves. And in reverse: that is. the multiple points are signal sources, and the single point you're trying to identify is the only receiver.

Did you know that if they didn't take Einsteinian time dilation into account (two sources -- receiver being in a gravity well and sources accelerating through space), then GPS calculations would lose accuracy by 10 feet a day, until they became completely useless in a year or two.
Courk
Posted: Mon Jul 04, 2011 4:19 am    Post subject: -4

Wouldn't you need 4 microphones? 3 points merely define a plane.
Antrax
Posted: Sun Jul 03, 2011 4:23 pm    Post subject: -5

I know why you asked, and I'm pretty sure FFT is what you need.
extropalopakettle
Posted: Sun Jul 03, 2011 4:08 pm    Post subject: -6

(What ever happened to Luna, and who knows why I asked?)
extropalopakettle
Posted: Sun Jul 03, 2011 4:05 pm    Post subject: -7

I was listening to an awesome thunderstorm today, and noticed in particular how the sound of thunder has this sort of two-dimensional texture, a single long rumble of it emanating different timbres and frequencies from different parts of the sky.

Now, I have a rough idea how human stereoscopic hearing works: The brain detects the latency between when a sound reaches one ear versus the other, and from this can judge it's position, at least along a one-dimensional left-right axis (we can usually get a good sense of up versus down too by tilting our heads, something we do unconsciously). We manage to do this for many sounds happening at once, locating them all in space.

Theoretically, with three microphones spaced apart, recording the same sound, one should be able to process it and calculate a relative direction from which the sound emanated.

I thought it would be cool to do this, and create visual representations of the frequencies emanating from different directions. This could be done with the sound of thunder, or a tree full of birds, or any setting in which many sounds are coming from different directions. The result should be a movie, a changing two dimensional field of color, with color mapped to frequency (there would be a loss of information, like timbre).

I'd imagine what I need is an algorithm which could take a sound file and convert it to data about the dominant frequencies over time. Anyone know what algorithms do this? Is Fourier transform what I want?

(I'm not looking for code, just the basic math / algorithms)