Skip navigation
Currently Being Moderated

I want to separate speech from an audio recording

Apr 30, 2012 6:36 AM



New to this forum.


I am about to buy CS6.


I have an audio recording which is an interview, but with some unwanted audio in the background.


Is there a tool available to extract only  the speech from the recording to effectively throw away everything else


Thanks for your help in advance


Chris Anderson

  • Currently Being Moderated
    Apr 30, 2012 8:37 AM   in reply to cjacs987654

    Well, we're all waiting for a chance to try CS6 but, even sight unseen, I'd bet my pension that the answer is "no".


    There will be various tools that you can use to enhance your voice and reduce the background (exactly which work best for you will depend on what sort of unwanted noise it is and how severe the problem) but it will be pretty much impossible to work as you want.  Some of this will also depend on whether your recording is stereo or mono and how it was recorded. A simple 60 cycle hum can be pretty much eliminated, tape hiss you can do a good job on--but music playing next door is probably there forever.


    Your best bet will be to experiment a bit and possibly to post a sample here for advice.

    Mark as:
  • SteveG(AudioMasters)
    5,602 posts
    Oct 26, 2006
    Currently Being Moderated
    Apr 30, 2012 1:41 PM   in reply to cjacs987654

    The most sucessful way of doing this only works with stereo files, where the voice you want is in one place, and all the interference is elsewhere in the stereo field - there's a tool for getting rid of that, and that's been present for the last several versions. But with mono, you often run into the significant problem of the wanted and unwanted sounds being in the same frequency range, and on the same single track. Remove part of the noise, and you unavoidably also remove part of the signal. Usually when you do this, the side effects are worse than the interference on the original signal. With wind, you stand some degree of chance of removing some of the effects - the low frequency rumble it creates is outside the speech band, and can easily be attenuated to quite a degree using EQ. But the rest of it is a lot harder - using downwards expansion sometimes helps a little if the wind noise is relatively quiet, but here you really are in the realms of potentially making it worse rather than better.


    You might find that using adaptive noise reduction helps a little - but without actually trying it,  it's very hard to tell.

    Mark as:
  • Currently Being Moderated
    May 8, 2012 2:56 PM   in reply to cjacs987654

    Here is a video tutorial by Colin:



    It will apply to CS6 equally.



    Mark as:

More Like This

  • Retrieving data ...

Bookmarked By (0)

Answers + Points = Status

  • 10 points awarded for Correct Answers
  • 5 points awarded for Helpful Answers
  • 10,000+ points
  • 1,001-10,000 points
  • 501-1,000 points
  • 5-500 points