Keep voice, remove background noise and music

[Scott Smolin]

In the past I used stereo equalizers to boost the mid frequencies that vocals are mostly on, and cut the high and low frequencies that noise and music are mostly on, of course this will not completely remove the noise and music just boost the vocals way above the noise and most of the music, with a PC usually theres vocal (for singing)
or speech(for speaking)preset EQ ssttings that do esentially the same thing.

Anyway this is probably the best that can be done since some of the music is actually in the same frequency range of the vocals you can’t completely remove all the music without removing some of the vocals.

[John Murrell]

You need to combine a number of tools to achieve this. It will probably not be possible to completely serperate the vocal. I use MAGIX Audio Restoration which I use for general audio restoration https://www.magix.com/us/audio-cleaning-lab/. I often use its tools to widen the stereo image, seperate the centre and side channels even further. I then use bx_solo(free) https://www.brainworx-music.de/en/download?PHPSESSID=aeogopu3v7kbu0e0efe495ut54 or some other tool to ms decode the sound (seperate the mid channel from the stereo). Not a perfect solution.

[Ramtin Neydawood]

Hi John,
I have a very important audio file that I needs some cleaning up. Judging from your posts, you really know what you’re doing when it comes to this stuff. I was wondering if you would be willing to help me and I would gladly compensate you for your work. Please let me know…
Ramtin

[Tanja Taxinapraxi]

hy Rian, did you find the solution about background? i’m searching the same thing… i have one skype call registrated where is the music in backgroung and i want to delete that music so i can hear the voice and conversation… so, did you find the program or the way how to do it? i would be so grateful if you can help me too. thanks in advance. bye. tanja

[Ken Natan]

I found a solution to this. (worked for me at least)

Quite simple and straight forward!

Download (free open source!) Audacity.

Load your audio file. (Video file didnt work for me, i had to convert it to an audio before I could process the file, in my case it was a vob file)

Once loaded, select the area you want to filter/extract voice of. (Sample it first, then get excited … :))

Then go to Effects tab and scroll down until u see Low Pass filter. Play with the parameters and im sure you will have the solution there.

(I had bad noise on my wedding video which i wanted to get rid off and it seems to have done the job:))

Good luck

[Jv Diaz]

use center channel extractor.. keep the side levels at the lowest level maybe, and keep the center at below +10… but you can still here some background there…

[Roger Fevry]

Hello!

I am new here, so I wonder if anybody could help me solve my problem …
I have some nature documentaries with in the audio file:
– natural noises
– music
– a voice in Russian
_ and in the background the original voice in English

Now, my question is: could I remove the Russian (or lower it) and normalize the English?

Many thanks in advance …

Roger

[Lowell Niles]

Hi! In reply to the original post: I looked at it for about 2 minutes and figured out how to isolate the vocals.

First, apply the “remove vocals” effect. After this is done, select the entire waveform and do “capture noise print”. This sets everything EXCEPT vocals as noise. Then undo (command-Z on mac for example) the “remove vocals” effect so you’re back with the original file. The captured noise print is still there, so now all you have to do is “noise reduction (process)”.

Have a great day!

Lowell Niles
Creative Director, Sunword Studios
https://sunword.com

[Khan Roar]

Here is few of video links which might help you

https://www.youtube.com/watch?v=7gyGOGRIRbI

https://www.youtube.com/watch?v=dNW4bfqgsO0

https://www.youtube.com/watch?v=svjAfl1gcos

But before using these video techniques or tricks I would suggest try to upload video to your youtube account, and luckily it match with other video music then youtube will give you the option in copyright tab to remove the music below the video and also show you area where it match, press the remove song or music it will process the video viola the video has more music except vocals, it worked for me most of the time except few ones.

Good luck

P.S. I am newbie in audio extractor, so my comment is not 100% guarantee…

Some contents or functionalities here are not available due to your cookie preferences!

This happens because the functionality/content marked as “Google Youtube” uses cookies that you choosed to keep disabled. In order to view this content or use this functionality, please enable cookies: click here to open your cookie preferences.

[Ht Davis]

Does anybody else see it? If you can remove the vocals cleanly enough, output the file that way (“But but but…” STOW it and read on). Now you should have a file with vocal, and a file without. Split the tracks of both into mono. Phase one set and mix them together. At this point, your vocals should be back, with some noise left.
I’d suspect your problem has to do with mic distance, how hot, type of mic and\or how loud the person was speaking\singing. When this kind of thing happens, those are usually the culprits. When you have to fix it, note the following:
Is the noise as loud as your speaker or louder? Remove vocals, then phase to bring them back and work from there.
If the noise is softer than the vocals, option 1 works, but you can also resignal the whole thing. It’s frowned upon in some circles, but it’s saved my butt in several instances. ReSignal is an old cleanup method to get rid of amp noise or process noise. It works because it is an ELECTRICAL process, not a digital one.
Resignal makes use of the following standards for measure:
75% on any knob for volume is considered a 0 or no amplification, any above attenuates down, any above attenuates up, with most knobs attenuating +6 or +10 at 100%.
Most equipment will produce some noise when amplifying a signal, but not when de-amplifying it. Some equipment however, will still apply a resistance noise.

You’ll need to know the zero points on your input equipment, have 2 DAW stations set up or 1 and another recorder with output capability. Resignalling requires enough time to play back the signal multiple times, and enough hard drive space to hold each pass multiplied by how many times you route back and forth. Explanation: To resignal, you’ll output from one station, through an output card or box (xlr or TRS), something with a knob attenuated down from 75% to a bit lower, like 65 or 60%, out to the input of the other computer, attenuated to 75% or 0 adjustment. This track will output to the first computer again, from a device (amp IO box interface or mixer console) set at similar adjustment below 75%, with the first computer set the same way on the input (0 adjust). Most IO boxes have several inputs and outputs. IF you have stereo audio, you’ll need 2 inputs for each send. This attenuates the actual signal you record and passes it back out at a slightly lower level. This smooths out the fundamentals and basic shake of the noise, so long as it’s below the good audio far enough (a few dbv). The vocals can get to a point barely audible, and still produce enough signal to be useable. Just remember to offset the original track so it gives about 12-15 seconds of dead air you can use for noise printing. What to watch for: IF you can play the original or a a slightly lower volume track through a set of headphones while you are recording a pass, watch the meters on the other tracks (which should be connected to your other inputs)–the meter should move only when your vocals pass in or when the noise is just loud enough (you can attenuate by a smaller amount with a final pass), and the difficulty will be watching and working 2 screens at once. Once the noise is at a low enough level, we’ll clean off the noise. Gather the 12-15 second front dead air, and combine it all into one sound, in its own track, save it to a file, open it up, select it, use the noise reduction plugin, capture and save the noise print. Open your final audio track you chose in file mode (not in multitrack editor anymore), run the noise reduction on it, but keep it between 80 and 95%. Now reverse the process you just did by amping the knobs by only a small amount, and use your new deamplified file as the starting point. Repeat the noise reduction process like before. Then do a new noise print on any remaining background noise. DOn’t ask why, I just know it works.