VideoHelp Forum

Our website is made possible by displaying online advertisements to our visitors. Consider supporting us by disable your adblocker or Try ConvertXtoDVD and convert all your movies to DVD. Free trial ! :)
+ Reply to Thread
Results 1 to 6 of 6
  1. Hi guys (and gals)

    My question is similar to that posed by other posters here in regard to removing background music so as to only hear the foreground voice or voices - except that here I need to reduce but not eliminate a lot of crowd chatter.

    The video / sound was recorded by someone else (honestly!) at a conference in Scotland (as you'll be able to tell from the accents in the clip at the end of the link below) and contains different speakers in sequence. The level of background chatter is also not consistent. Some segments are very noisy and some quite acceptable. I was asked to edit the clips recorded into a five minute presentation for use by the charity referenced in the sequence.

    Their feedback was that they found the levels of background 'babble' intrusive and, although the video has text subtitles for what the speakers are saying, hard to hear what the speakers are saying.

    I know that this sort of background noise - by virtue of the fact that it shares many of the frequencies and amplitudes of the foreground waveforms - is very difficult to remove or reduce without quite a big hit to the bit of the speech you want to keep. I have tried using Audacity to 'tidy up' the soundtrack (using its 'noise reduction', high and low pass filters and the compressor as suggested elsewhere) with limited results. Is there anything else that could be tried? Is someone 'out there' willing to have a go to see if they can do a better job than I was able to? (The second link will be to the edited soundtrack results).

    My thoughts, such as they are, is that it might be possible, with the 'right' software, to differentiate the spoken words from the chatter by software (or an add-in to an existing app) which has an algorithm to detect spoken words or which can be given the waveforms of the sentences spoken by the main speakers and to use this to try to recognise and extract those waveforms from the flow of sound and, perhaps, to put it to a separate channel or even file...

    I realise that this is not likely to be a 'highly precise' method - given the vagaries of speech such as speed, accent, legibility etc, but at least if this software had an ability to be 'trained' or parameters which could be adjusted, perhaps...??

    Does such software exist? If so - does anyone have any experience of it? Can it be dowloaded for free (unlikely, I know!) or paid for? Or is there an agency which undertakes such 'repairs' - for a fee? (or, indeed if there is someone out there who might be able to do this with a greater degree of success than me we could discuss possible fees - which wouldn't be huge as it is for a charity so they don't have deep pockets... And where I say 'success' I mean where the speaker keeps most (if not all!) of their speech frequencies and the background chatter is very much reduced - but NOT eliminated - to the levels such as at the start of the sound file...

    Many thanks for your invaluable help.

    JC from the UK.

    Link to both files:

    First file (Heart of Scotland audio only unmodified): Original audio file with unadjusted sound

    Second file (Heart of Scotland audio only no music V2 adjusted): with adjustments I made with Audacity:
    Quote Quote  
  2. Member ChapmanDolly's Avatar
    Join Date
    Sep 2017
    United Kingdom
    Search Comp PM
    The only software I can think of to tackle this kind of problem is iZotope RX6 which is a spectral audio restoration tool. It's not cheap though. The basic Elements version is around 157.00.
    Canon G40 - Dell XPS8700 i7 - Win 10 - 24gb RAM - GTX 1060 6GB - DaVinci Resolve 16.1b3 -- Dell dual Xeon Workstation - Win 7 Ultimate - 16gb ECC RAM - GTX 650 - Pro Tools 2018.3
    Quote Quote  
  3. Member
    Join Date
    Mar 2008
    United States
    Search Comp PM
    Yes, I've got the trial of the latest version RX7. It has a dialogue isolate module that helps in situations like these.
    Here's a sample
    Image Attached Files
    Quote Quote  
  4. Member
    Join Date
    Aug 2010
    San Francisco, California
    Search PM
    This is a good opportunity to remind people that you can't just toss a microphone in the approximate direction and expect good results. Fifteen minutes fiddling with mic placement can save hours of work later.
    Quote Quote  
  5. JVRaines - I entirely agree, which is why I spent quite a lot of time subsequently with the chap talking about mic positioning, different types of mic (Lavallier, shotgun etc) which could be used. I think he's learned (one of) the lessons!
    Quote Quote  
  6. Video Restorer lordsmurf's Avatar
    Join Date
    Jun 2003
    Search Comp PM
    I wonder what Sound Forge would do.
    Unfortunately, I'm just too busy to play around with this sample.
    Quote Quote  

Similar Threads