Connect with us

Subtract audio from noise

Discussion in 'Electronic Design' started by Edward Lang, Jul 10, 2007.

Scroll to continue with content
  1. Edward Lang

    Edward Lang Guest

    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
     
  2. 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.
     
  3. It can't really be done, in the analogue domain


    martin
     
  4. 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.
     
  5. Charles

    Charles Guest

    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.
     
  6. Phil Allison

    Phil Allison Guest


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

    Stop posting trolls.



    ....... Phil
     
  7. Robert Baer

    Robert Baer Guest

    There is a *MUCH* simpler way to generate noise...
     
  8. Rich Grise

    Rich Grise Guest

    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
     
  9. whit3rd

    whit3rd Guest

    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.
     
  10. Eeyore

    Eeyore Guest

    To *subtract* the audio ? You mean you just want the noise signal ?

    Most ppl want to subtract the noise from the audio.

    Graham
     
  11. 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.
     
  12. mpm

    mpm Guest

    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
     
  13. Al

    Al Guest

    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
     
  14. Eeyore

    Eeyore Guest

    It won't I'm afraid. There will be some cancellation but it'll be far from
    complete as the 2 mics won't pick up the 'audience' in an identical manner.

    Graham
     
  15. john jardine

    john jardine Guest

    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 :)
     
  16. Don Bowey

    Don Bowey Guest

    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.
     
  17. 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.
     

  18. 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.
     
  19. Don Bowey

    Don Bowey Guest

    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.
     
  20. J.A. Legris

    J.A. Legris Guest

    Ah, no:

    http://www.bose.com/controller?event=view_product_page_event&product=qc2_headphones_index
     
Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day

-