Q: How to Remove background music or sounds

[default]

Hi everybody,

I know this is a difficult to find a solution situation cause I already searched the forum but if you don't ask you never find out. I am having some videos (mpegs / avis /wmvs) where the background music is so lame that I wish I could find a way to either totally erase it while leaving the movie dialogue intact or at least reduce the volume level and surpress it. In a similar situation the background noice -coming from a running water tub- is making so much noice that the dialogue is nearly impossible to hear.
If there is a solution through re-encoding with either virtual dub and using a filter or with any other software I would be very greatfull to hear it.

[setarip_old]

Hi!

If I remember correctly, the (freeware) program "Audacity" inlcudes a tool that allows you to select a small portion of the audiotrack and identify the offending noise value in it - and then seek and remove it from the entire track...

Other audio editing programs, such as GoldWave, have noise filters available, as well...

[jordisound]

Using audacity you can remove noise, but what you can't do is separate music and voice if are both in a track. With 5.1 you can select only the channel you want, but you never will be able to separate music and voice in a stereo.

Use of audacity is simple. Select a part of track when only you hear noise (in your case background music) and use filter denoising. I don't think it works ok.
regards

[default]

Thank you for your answers. I have installed audacity and although it works decently on mp3 files it doesn't work on video files. It loads them and while I am trying to play them so that I will find the point where I can take the audio sample that I can use as a filter it sounds like it plays them in a non recognizable way. Are you sure audacity works for video files? Is there any otrher software I can use alternatively for video?

[Mug Funky]

sorry to say it, but there's no chance in hell, purgatory or heaven that you're going to be able to get the music out, unless it's multichannel audio mixed such that the dialogue is on one channel and the music on others (like music on left and right, and dialogue on centre, with effects on left right and surrounds). this isn't likely to happen with mp3 or wma (both are typically stereo).

FFT based noise reduction will not work on music - it's effectively a passive filter, where music is very much dynamic. as soon as the music differs from a rhythmless, tuneless monotone you'll hear uglies coming through.

sorry to bear bad news :)

[jordisound]

Are you sure audacity works for video files?
Every audio editors works with .wav files. It doesn't matter the .wav is from music, video, etc.
Audacity can't works with AC3, you have to convert .wav

[Soundfx4]

sorry to say it, but there's no chance in hell, purgatory or heaven that you're going to be able to get the music out, unless it's multichannel audio mixed such that the dialogue is on one channel and the music on others (like music on left and right, and dialogue on centre, with effects on left right and surrounds). this isn't likely to happen with mp3 or wma (both are typically stereo).

FFT based noise reduction will not work on music - it's effectively a passive filter, where music is very much dynamic. as soon as the music differs from a rhythmless, tuneless monotone you'll hear uglies coming through.

sorry to bear bad news :)

I know this is old, but I just happened upon it on google, and decided to share my thoughts.

While I'm not 100% sure how it could be done, I know that (at least theoretically) it should be able to be done, although definitely not automatically; instead it'd be done by a professional using an audio editing program with lots of filters and tools. My reasoning be that the vocals and music operate at different frequencies, and as long as they never cross, a professional should be able to separate them in many cases. Obviously though some songs would be more difficult than others.

And then again, I'm just guessing that this can be done based on what I know about sound, but at the very least, I can say without a doubt that, theoretically, it can be done...but that doesn't really mean anything when it comes down to it.

[JoeShrubbery]

You know I'd always wondered if it might be reasonably possible to do a sort of inverse voice removal.

Many karaoke systems do simple "voice removal" on existing stereo tracks by phase-shifting one of the channels then downmixing to mono. This works because the vocals are typically already centered across the stereo signal (effectively mono), so the originally identical frequencies in the left and right channels, after being phase-shifted and merged together, cancel each other out thus effectively removing those frequencies from what you or I hear.

I've wondered for a long time now if there might be a way to isolate those identical frequencies between the two stereo channels, basically identifying what would be cancelled out in a simply voice removal system and instead keeping them while discarding everything else. Logically I'd think it's doable, I just lack the programming skills to come up with something to try it out.

[raquete]

@ default
*read my next post first!!!

where the background music is so lame that I wish I could find a way to either totally erase it while leaving the movie dialogue intact or at least reduce the volume level and surpress it.

if is .wav,ac3 or mp3,can you post(rapidshare) 1 minute of sample(audio only) to try my tests?

[raquete]

If there is a solution through re-encoding with either virtual dub and using a filter...

have a "center cut" filter that can cut sides and center in vdub,try it first!

[Anatoli]

What about removing background music from MONO files? No chance?!

I've got the feeling that this complicated task can be solved using the following steps:
- Pattern analysis and editing for voice recognition;
- Determination of regions using voice recognition;
- Lowering of volume on the entire track except the voice regions

Easy to say, hard to implement though... Any ideas how?

[raquete]

What about removing background music from MONO files? No chance?!...maybe

Any ideas how?to try i need to know your souce.
can you upload in rapidshare few Mbs of your audio to do my tests here?

[Anatoli]

Any mono wav file will do. Just convert stereo into mono (only left or right channel) and start with sample material where voice is easily distinguishable.

It is important to solve the problem in principle. To me, automatic recognition of voice regions in a sound track is the hardest part.

I was thinking about using center channels from SACD or DVD-Audio as a pattern source for certain artists.