VideoHelp Forum
+ Reply to Thread
Results 1 to 6 of 6
Thread
  1. Member
    Join Date
    May 2007
    Location
    United Kingdom
    Search Comp PM
    I'm trying to get the music from a documentary which has various languages in the audio options, so the narrator's voice is separate from the music, but the voice plays over the music on both speakers. Is there a way to separate the two?

    Here is the audio that I mean


    http://www.zippyvideos.com/3495098356964916/audio/





    ---------------------------------------------------
    Quote Quote  
  2. Always Watching guns1inger's Avatar
    Join Date
    Apr 2004
    Location
    Miskatonic U
    Search Comp PM
    The narrators voice isn't separate, they have simply recorded different narrations over the same music.

    You could try something like AnalogX's VocalRemover, which is free, however there is a simple caveat that goes with it (and pretty much all vocal removers). The vocals must be recorded equally in both channels (i.e. straight down the centre of the stereo mix), and the music must be well separated in the mix. If the vocals are down the centre, they won't be removed. If the music, or any part of it, is equally mixed in the left and right channels, it will also be removed.

    Finally, if you listen to the music, you will notice that when the narration is audible, the music dulls in tone and volume to accommodate it. If you did manage to remove the vocals, you will still have to adjust the music to restore these areas.

    In short - lots of work with little likelihood of success.
    Read my blog here.
    Quote Quote  
  3. Member
    Join Date
    Dec 2005
    Location
    United States
    Search Comp PM
    hi,
    first i am no expert.. but i try.. smile...
    I agree with gunslinger, if the have mixed the music and narration together before they brought it into the video.. then extract the narration from the music is going to be very difficult and in a lot of cases don't work well... I have a couple of programs... one came with my buring suite and the other audacity... it a toss of the coin on how well the extraction is... but definitely need to make sure first you have your dc offset done... then go thru procedures...

    however... if what they did, in the video file they put the music on one track and they put the narration on a different track then that maybe easy... you bring up the video in yur movie editor.. let say window movie maker and see if they did, if they did then just a matter of deleting the mic naration....



    Originally Posted by Aaron1927
    I'm trying to get the music from a documentary which has various languages in the audio options, so the narrator's voice is separate from the music, but the voice plays over the music on both speakers. Is there a way to separate the two?

    Here is the audio that I mean


    http://www.zippyvideos.com/3495098356964916/audio/





    ---------------------------------------------------
    Quote Quote  
  4. Yeah I would try Audacity(Effect->Noise Removal) and filter out the voice,then load the file again and filter out the music.
    As others have said the sound will be distorted and is not perfect.
    Here's a sample:


    voice.mp3
    Quote Quote  
  5. Member
    Join Date
    Aug 2005
    Location
    England
    Search Comp PM
    Hi there, I had a go with Wavelab 5.1. I don't EQ as a rule, being an Audiophile I normaly spit at such things. I think I go for Compression to level everything equal, sounds strange but like the other guy said, the music fades a little to allow clarity of speech. Next, you'd need more than likely, a Parametric Equaliser, preferably one that allows several filters to stack, I think a VST Plugin called Equilibrium is one, somthing like that. Google for it. The purpose of this over a Graphic Equaliser is, you can create a filter as simle as a tone control, through to shelving(highpass/lowpass), Band pass and Notch. You can vary, say the Notch. Look at a typical Graphic EQ, say mid band is 1khz, what if you want 999hz? or 1,001hz? A parametric could help shift the speech out. If the speech is dead centre you could try Panning to shift it/widen the soundstage. You could try a bandpass, make a copied file.Leave one normal, bandpass on the other to remove the music leaving the vocal, Invert that file and paste it back over the first file, if all is well, the vocal/speech would get cancelled out. Try wearing a decent pair of headphones, not the inner ear Ipod jobs, but a full size pair of cans.
    Beware any background noise masking things, hope you got a good source/recording, Wavelab, Soundforge or Audacity and patience. It aint easy and won't be perfect. Good way to train your ears too.

    Mike
    Quote Quote  
  6. I've done this a while ago with a documentary with a very bad translation.
    If your software have the possibility to "keep" only the centered audio (I used a freeware dsp effect).
    Take the left channel of two of the different languages, make a new stereo file containing one language in left and another in the right channel and remove the stereo content. then do the same for right channels. Finally make a new stereo track from the "fixed" left and right channels.
    these steps gave me a very good sounding track with no detectable distortions.
    Quote Quote  



Similar Threads

Visit our sponsor! Try DVDFab and backup Blu-rays!