Separating Sound Effects, Dialogue and Music?

[JCDragonBall]

You're creating a lot of threads. Try to check there aren't already existing relevant threads you can add to before creating a new one - there's definitely a few covering this topic already, and we even have one dedicated to "random thoughts".

Edit: in fact, threads merged.

Apologies! Thank you

Try Spleeter, linked above.

I will try this out! Thank you!

 

[Fran_Garcia]

There is a new plug-in from Waves named Clarity VX. I watched a review and then tried it by myself, and I have to say that it is a great tool for removing background music, FXs and isolating the voices or dialogues. I do recommend it.

Check it out:

Clarity Vx – Clean Vocals. Fast. No Compromise. | Waves

Clarity Vx is the highest-quality AND the fastest way to clean your vocals from background noise and prepare them for any mix, production, podcast or video. Powered by Waves Neural Networks.

www.waves.com

 

[DigModiFicaTion]

I'm getting pretty excellent results on isolating vocals/dialogue using https://mvsep.com/ and the MDX 2 setting (see the chart on the page for quality scores to identify what settings are recommended). It does have a 10 minute track length limitation, but it's easy enough to break a surround file up into 10min portions using audacity.

 

[lantern51]

I have found this site useful for just separating dialog:

melody ml

Melody ML lets you easily seperate audio tracks using Machine Learning. Automatically isolate vocals and generate stems to remix songs with.

melody.ml

 

[TM2YC]

I have found this site useful for just separating dialog:

melody ml

Melody ML lets you easily seperate audio tracks using Machine Learning. Automatically isolate vocals and generate stems to remix songs with.

melody.ml

Wow just did a test on an 8-second clip of movie audio with music and FX all over it and it instantly came back with pretty clean, and minimally distorted vocals. The removed sound FX and music is separated onto other tracks, that while not clean of vocals, could certainly be useful to adding back sound FX if mixed correctly.

I'll be investigating this some more. Are their upload limits, or limits on length of audio tracks, which you've encountered?

 

[lantern51]

Wow just did a test on an 8-second clip of movie audio with music and FX all over it and it instantly came back with pretty clean, and minimally distorted vocals. The removed sound FX and music is separated onto other tracks, that while not clean of vocals, could certainly be useful to adding back sound FX if mixed correctly.

I'll be investigating this some more. Are their upload limits, or limits on length of audio tracks, which you've encountered?

I haven't come across anything limiting me as of yet, but the scenes I've used are short. About a minute and a half was the longest one. I see where they say something about credits remaining but every time I go back to the site I always have the same 2 credits. I don't know if that is the site or my browser being one of the "safe browsers"? I have even uploaded movie clips instead of sound files and it worked.

 

[GelflingHand]

I'll be investigating this some more. Are their upload limits, or limits on length of audio tracks, which you've encountered?

The FAQ says max size 100Mb. It's based on Spleeter, which I didn't have much luck with because I was trying to separate singing from dialogue, not just music. If they start enforcing the two song limit, there are other Spleeter-based web services, or you can install Spleeter on your own computer (it can be a little tricky to get working, but not too bad). I'd be curious whether you get noticeably different results from the various Spleeter web services.

If you just want the vocals, not preserving sound FX and music, I've had some good results with https://audo.ai/noise-removal

 

[DigModiFicaTion]

Getting some really clean vocal only tracks using Ultimate Vocal Remover 5. https://github.com/Anjok07/ultimatevocalremovergui
Pretty impressive isolation. I believe you have to prune down your audio files to 10 minutes per pop, but the conversion process is pretty fast. I drop the mkv into audacity, grab the center channel and export as a wav (full for a base) and then 10 minute increments and load them one by one into UVR5. Shoots out a wav file that I load into Vegas.

Update, I was able to do a vocal isolation on a 2 hour wav file! No need to create smaller files

 

[asterixsmeagol]

I may give this a shot for some stuff on my Picard edit. The dialogue bleeds into the front left and right channels almost 100% of the time, which is making some audio transitions tricky.