Kuro5hin.org: technology and culture, from the trenches
create account | help/FAQ | contact | links | search | IRC | site news
[ Everything | Diaries | Technology | Science | Culture | Politics | Media | News | Internet | Op-Ed | Fiction | Meta | MLP ]
We need your support: buy an ad | premium membership

Can DSPs Save Us From Cell Phone Noise?

By kipster in MLP
Tue Feb 20, 2001 at 08:33:55 PM EST
Tags: Technology (all tags)

I've always wanted to have ears like a dolphin. Here's an interesting invention-disclosure document (anti-patent) that suggests that we can use DSP chips in cell phones to do some echo location and find the position of the speaker. Then we just filter out the noise that comes from a different location.

I can think of hundreds of good applications for some good open-source DSP code. If I only knew something about the area, I would start a SourceForge project to develop some good libraries and toolkits. Today's generic CPUs have more floating-point firepower than the DSPs from several generations ago. We should be able to do way cool things with built-in echo location and position discrimination.

My favorite idea is to have some sort of voice recognition tool that identifies the speaker and then translates it accordingly. You might be able to do this without echo location, but it would still be neat.


Voxel dot net
o Managed Hosting
o VoxCAST Content Delivery
o Raw Infrastructure


I Would Use Sound Postion To
o Fix That Rattle In My Car 17%
o Search for Interesting Birds In My Backyard By Their Call 12%
o Eavesdrop on One Table In the Lunchroom 15%
o Figure Out Which Punk Keeps Coughing "asshole" In My Class 17%
o Get A Cool Speakerphone 7%
o Figure Out Who Uses the Word "Sex" Often In Bar Conversation 30%

Votes: 40
Results | Other Polls

Related Links
o invention- disclosure
o SourceForg e
o Also by kipster

Display: Sort:
Can DSPs Save Us From Cell Phone Noise? | 17 comments (12 topical, 5 editorial, 0 hidden)
DSPs (3.00 / 1) (#1)
by Bad Harmony on Tue Feb 20, 2001 at 10:32:21 AM EST

DSPs are much cheaper and more power efficient than general purpose CPUs. Plus, you don't have a general purpose operating system that does nothing but get in the way.

You can do phasing tricks with radio frequency antennae and sonar arrays to get directional patterns. I don't think this will work with speech for several reasons. The frequency range (300-3000 Hz) is a full octave, way too large for phasing to work. The wavelengths are so long (78 cm at 440 Hz) that the microphones would have to be spaced far apart.

5440' or Fight!

Out of Curiosity (5.00 / 1) (#6)
by reshippie on Tue Feb 20, 2001 at 11:48:52 AM EST

The frequency range (300-3000 Hz) is a full octave, way too large for phasing to work.

Did you mean 300-3000Hz, or an octave? I don't mean to nitpick, but an octave above 300Hz is 600Hz.
That means the difference between 300Hz and 3000Hz is about 3 octaves, plus a minor 3rd (I think).

Those who don't know me, probably shouldn't trust me. Those who do DEFINITELY shouldn't trust me. :-)
[ Parent ]

Octaves (3.00 / 1) (#7)
by Bad Harmony on Tue Feb 20, 2001 at 12:05:21 PM EST

You are correct, I was confusing octaves with orders of magnitude.

5440' or Fight!
[ Parent ]

Fair enough. (NC) (2.00 / 1) (#14)
by reshippie on Tue Feb 20, 2001 at 04:08:15 PM EST

No Comment.

Those who don't know me, probably shouldn't trust me. Those who do DEFINITELY shouldn't trust me. :-)
[ Parent ]
Microphone arrays (none / 0) (#16)
by galibert on Tue Feb 20, 2001 at 08:52:07 PM EST

Take 59 microphones in a line each 1.5cm apart and you'll find out that phasing (well, beam forming is the technical term) works quite nicely for voice at a distance in a noisy environment. Of course, the 1m size of the array is not practical for a cellphone :-)


[ Parent ]
Noise (3.00 / 1) (#2)
by wiredog on Tue Feb 20, 2001 at 10:34:27 AM EST

The patent application and article talk about ambient noise. When I use a phone I find the noise in the signal (i.e. static) to be more annoying. And that can't be filtered out. Mandelbrot was researching that when he found the Mandelbrot Set. Noise is fractal. No matter how well you filter, boost signal strength, or whatever, there is always some noise. The math can be hairy, and I don't have it in front of me anyway. Besides, I don't think scoop supports MathML.

The idea of a global village is wrong, it's more like a gazillion pub bars.

cell phone noise (4.00 / 1) (#3)
by mstevens on Tue Feb 20, 2001 at 10:39:39 AM EST

Most of the world has solved the problem of noise on cell phone calls by going to digital cellphones...

[ Parent ]
noise (none / 0) (#4)
by mstevens on Tue Feb 20, 2001 at 10:44:18 AM EST

Obviously, this only deals with noise in the signal.

[ Parent ]
Intelegent Filtering (4.00 / 1) (#5)
by delmoi on Tue Feb 20, 2001 at 10:53:53 AM EST

Sorry, I'm away from a spellchecker right now

First of all, you're not really using echo-location there, echo-location is when you send out 'clicks' and listen for their echo, basicaly a way of using using sound waves the way our eyes use lightwaves. You might as well just stick cameras on the phone if you want to to know where someone is.

What the invention seems to be saying is that you can use extra microphones, but as someone pointed out, the wavelenghths are pretty big, so that might not work.

I've actualy been thinking about a way to get rid of cell-phone noise, and I think I've found a solution. Does anyone remember those 11-node nural nets that were mentioned on slashdot a while back? it could do things like pick a voice out of static that was something like a thousand times louder, or have a recognition accuracy of 60% when *other* voices were 500 times louder. Well, assuming you could use this thing to extract the target waveform, I think it would be ideal for things like cellphones. You wouldn't even need to have the mic near your mouth (so you could use an earphone type thing).

And as far as line static, digital cellphones already get rid of most of that. Well, they at least replace static with random packet drops....
"'argumentation' is not a word, idiot." -- thelizman
DSP and voice recognition (none / 0) (#15)
by nickp on Tue Feb 20, 2001 at 07:06:46 PM EST

First of all, DSPs are usually more powerful than general processors in the area they are designed for: digital signal processing. They usually have instructions that allow for lots of parallel FPU commands, e.g. vector math. DSPs are not a thing of the past. As far as open-source DSP code, the specs for many DSPs are publicly available, e.g. Texas Instruments has instruction listings, etc., so it seems feasible.

I don't see how echo location and position discrimination would be relevant for speech recognition. Speaker-independence is very important for any good speech recognition package. Echo location doesn't help us recognize people's speech better. It will only tell us whose speech we are recognizing, since you still would need to know a person's invididual phonotactics to enable appropriate recognition based on speaker identification. But then speech alone should be sufficient to enable appropriate recognition as well.

"Gravitation cannot be held responsible for people falling in love." -- Albert Einstein

Background noise? (none / 0) (#18)
by jovlinger on Wed Feb 21, 2001 at 01:01:58 PM EST

It was my understanding (poor, in general) that one of the biggest hurdles for decent voice recognition was background noise. If position filtering allows us to strip away 90% of background noise, it is likely that accuracy rates will improve accordingly.

but I have to confess, I'm guessing here

[ Parent ]
DSPs to save us from cellphone noise eh? (none / 0) (#19)
by DJNW on Thu Feb 22, 2001 at 05:52:12 PM EST

now that there's a way to save someone using a cellphone from the noise of everyone around, perhaps we can get back to work on a way to save everyone around from the noise of cellphone users...

Can DSPs Save Us From Cell Phone Noise? | 17 comments (12 topical, 5 editorial, 0 hidden)
Display: Sort:


All trademarks and copyrights on this page are owned by their respective companies. The Rest 2000 - Present Kuro5hin.org Inc.
See our legalese page for copyright policies. Please also read our Privacy Policy.
Kuro5hin.org is powered by Free Software, including Apache, Perl, and Linux, The Scoop Engine that runs this site is freely available, under the terms of the GPL.
Need some help? Email help@kuro5hin.org.
My heart's the long stairs.

Powered by Scoop create account | help/FAQ | mission | links | search | IRC | YOU choose the stories!