Subtract audio from noise

What would be the simplest analog (not DSDP) way to subtract a complex audio signal, such as voice or music, from broadband white noise?

Thank you for any advice.

Ed Lang

Reply to
Edward Lang
Loading thread data ...

Last century it was done by CQ-DX'ers by "properly" cleaning ears. Not sure what is the "new, improved, updated...." formula for new millenium. A ticket to La Scala maybe?

Have fun

Stanislaw.

Reply to
Stanislaw Flatto

It can't really be done, in the analogue domain

martin

Reply to
martin griffith

What do you mean "subtract"? In the mathematical sense, point by point?

There's not much point in doing so, as audio generally has undefined and irrelevant phase.

Which means it makes little difference if you subtract or add.

And if you add, well addition is commutative, so you're adding noise to audio, which you can do with two resistors.

--
But then we\'re probably misunderstanding you.  Maybe it would help if
you\'d amplify your explanation, or explain why you want to do this, or
 Click to see the full signature
Reply to
Ancient_Hacker

A phase inverted signal can be used in RF communication when two antennas are available. However, I think that's limited to a specific noise source (lightning, e.g.) ... not white noise.

Reply to
Charles

"Edward Lang"

** Do your own Google on " audio noise reduction ".

Stop posting trolls.

...... Phil

Reply to
Phil Allison

There is a *MUCH* simpler way to generate noise...

Reply to
Robert Baer

By "subtract", do you mean "extract"? In that case, you're out of my league; but if you really mean "subtract", then just a subtractor circuit.

Good Luck! Rich

Reply to
Rich Grise

Filtering, of course. Select the parts of the bandwidth that carry the largest voice/music information, and attenuate everything else.

Or if it was the noise you wanted, attenuate the voice/music.

There are other answers for SIMPLE audio signals, which involve making a model of the audio signal (if it's repetitive or otherwise predictable, this can work). For a sine wave, for instance, a phase-locked loop can reproduce the signal even after adding LOTS of noise. And Shannon's Theorem (on information content of noisy signals) tells you that it won't be possible to remove any signal of more complexity than the information, after noise is taken into account, of the signal+noise.

Reply to
whit3rd

Why do you need to subtract the audio? What is special about the white noise you are going to finish up with?

Most people want to* extract* audio from white noise.

--
~ Adrian Tuddenham ~
(Remove the ".invalid"s and add ".co.uk" to reply)
 Click to see the full signature
Reply to
Adrian Tuddenham

I'll have to refresh myself on the spectrum distribution of "white" noise, but my first thought would be to truncate everything above say... 10kHz. That should get rid of a bunch of it, albeit along with the high end desired audio signal as well.

Are you building hardware, or just need the job done? There are several PC-based software applications that can do this, if that approach would work for whatever you are doing. Cool Edit Pro (Now known by some other name since Adobe bought them) comes to mind.

But I agree with the other poster, this really can't be done (to my knowledge) without digital sampling.

-mpm

Reply to
mpm

If your are recording, say a choir, and want to remove the noise from the audience, do this. Point one microphone at the audience and one at the choir. Invert the signal from the audience and add it to the one from the choir. Adust the amplitude of the audience microphone so that it just cancles the audience sound from the choir microphone.

Of course, you may want to use multiple microphones and point them to maximize your cancellation.

Al

Reply to
Al

Normally it would be a opamp wired as a difference amp. But the question is interesting in that it would imply that white noise intrinsically contains all the sounds ever produced since the universe began, or liable to be produced till the end of time. If this is the case, then subtracting the particular audio signal from this cacophony will leave noise still containing an infinity of audio signals but minus 1 of them. That is, it'll still be white noise :)

--
Posted via a free Usenet account from http://www.teranews.com
Reply to
john jardine

But if the white noise is present in real time - not a recording - then it is possible to receive the noise on a clear frequency, amplify it, invert it and sum it to leave the wanted signal. It is done effectively in better communications receivers.

Reply to
Don Bowey

Well, it does, in a simpleminded theoretical sense.

But in the real world sound and microphones have wildly varying frequency response and phase, which makes this method worse than useless.

Reply to
Ancient_Hacker

Ah, no. Noise by definition is not predictable. The noise at 1000KHz does not correlate well with the noise at 1001KHz. Even if it did, you'd need two receivers with impossibly matched frequency, bandwidth, and phase response.

There are "Noise blankers", which work great, but only for your basic Harley motorcycle spark plug impulse noise. Which isnt really random noise, but has a very regular and predictable structure. That's why it's possible to do.

Reply to
Ancient_Hacker

Why do you you think a simple detect/invert/sum analog process needs "very regular and predictably structured noise?"

The ones I've used work well with random noise.

Reply to
Don Bowey

but

Ah, no:

formatting link

-- Joe

Reply to
J.A. Legris

If you add two random noises, you just get more random noise.

By definition, random isnt correlated with anything else, especially other random sequences.

Without exact algebraic correlation, you can't subtract and get a reduction.

Now you can clip, divide, or blank impulse noise, but that's not noise in the hissy sense. It's reliably short and spiky and above the desired signal level, making it reliably detectable. You can't do the same for hissy noise, by definition. If you could, then we'd be able to pick up several million TV stations from other galaxies with just a tea saucer covered with tn foil.

Reply to
Ancient_Hacker

And if both random noises are from the same source at the same time?

Certainly it correlates to itself......

Atmospheric noise, in which I include galactic white noise, covers a wide spectrum, and what is detected at frequency A will also be detected at frequency B.

If I am listening to a radio station at frequency A (with atmospheric noise), and my "noise blanker" is tuned to a clear frequency (B), receiving the same atmospheric noise, it will work fine.

Reply to
Don Bowey

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.