Need Help with Audio Separation

[MarsEverythingTech]

Hello Adobe Audition community!

I hope you're all doing well. I'm currently working on a project involving an old VHS video that I've digitized on my PC. The video and audio are combined into one file, and my goal is to separate the vocals from the instruments in the audio. However, I'm encountering difficulties, particularly when using the audio after editing it in Audition on Moises App.

Here's a summary of my situation:

• I have an old VHS video recorded digitally on my PC.
• The video and audio are combined into one file.

My objectives:

1. Separate the vocals from the instruments in the audio file.
2. Edit the separated audio in Adobe Audition for further enhancement.
3. Utilize the edited audio in Moises App for additional processing.

Specific issue:

After editing the audio in Adobe Audition and using it in Moises App, the output doesn't meet my expectations. The separation quality seems to degrade, and the extracted instrumental sounds less clear than anticipated.

I'm reaching out to this supportive community to seek advice, guidance, or alternative methods for audio separation in Adobe Audition. Additionally, if any of you have experience with integrating Audition-edited audio into Moises App for further processing, I would greatly appreciate your insights.

If you've successfully accomplished similar tasks or have any knowledge of plugins, settings, or workflows that could aid me in achieving a high-quality audio separation, please share your expertise.

Thank you so much for your time and assistance. I'm eager to learn from your experiences and overcome this obstacle to create an impressive audio track from this old video.

Best regards, MarsEverythingTech

[Ares Hovhannesyan]

Here is some methods.

To separate audio vocals from sounds in Adobe Audition, you can use the following methods:

1. Work in a Multitrack editor, select where you want to split, select the track that you want to split and press Ctrl+K. Alternatively, click the right mouse button onto the track you want to split and in the menu click Split.
2. Use the Learn Sound Model and Sound Remover effects to remove instrumentals or vocals.
3. Use the Center Channel Extractor effect to create an a cappella track by removing instrumentals, or create a karaoke track by removing vocals.
4. Use the Amplitude Statistics window to highlight the noise floor, click Scan Selection and make a no...¹

1. voiceovertip.com2. youtube.com3. youtube.com4. youtube.com

[MarsEverythingTech]

Unfortunately, these methods didn't work. 

[SteveG_AudioMasters_]

The reason for that, I'm afraid, is twofold: Firstly, Audition doesn't have the tools to do this (it's a rather specialised tool you need to do it - something like Melodyne might help) but just as importantly, to do this at all (and even then it's not very good) you need a reasonable-quality original. An old VHS video does not remotely qualify.

Don't be fooled by the things you saw on CSI - they were all pretty much faked. What you are trying to achieve is like trying to unbake a fruit cake. You might get a few raisins out, but you are still going to be left with a heck of a lot of inseperable cake mix...

[MarsEverythingTech]

Hey SteveG_AudioMasters,

Thanks a ton for chiming in on my forum post about audio separation! Your insights about Adobe Audition's limitations and the challenges with the old VHS source were eye-opening.

I've been curious about exploring other options, and you mentioned Melodyne might be a better fit for this task. That sounds awesome, and I'd love to give it a shot!

If you've got some experience with Melodyne or know any tips, tricks, or tutorials to get me started, I'd be super grateful for your help. I'm eager to learn how to use Melodyne effectively to separate vocals and instruments from my combined MP4 file.

I totally get that the separation might not be perfect, but any improvement over the original would be fantastic.

Looking forward to hearing from you and soaking up your Melodyne wisdom!

Thanks again for your time and expertise!

[SteveG_AudioMasters_]

I'm afraid that I'm not a Melodyne expert, nor is it the only software that might be able to help - it was merely an example. I've yet to hear any separation software that does even a remotely acceptable job, even on good original material. Fortunately it's not something that I'd ever need to do; my strong inclination is to leave poor-quality material as it is. After a while you get to know intuitively what can be 'improved' and what can't - and virtually everything originating from domestic videotape falls into the latter category.

One of the significant reasons for this is that the linear audio from domestic video recorders never has good pitch stability - this is a problem inherent in their mechanical design. If it's FM sound then it's slightly better, but that's pretty rare to find. Without good pitch stability any software attempting to isolate parts has an even worse struggle. So its inherently poor performance is made even worse. Oh, and if it's mono, then just give up - you stand no chance.

I realise that this may all sound rather negative but it is based entirely on experience and, as you will discover if you try to do any of it, rather realistic.

[Ares Hovhannesyan]

Please look on https://youtu.be/yuVwHp4iYKw

Here is method 2 that make very fantastic work. Inverting Music Chanel allow to seperate voice.

[Spamhunter]

Reminds me of a problem a client had with silent gaps in an audio file that had them flummoxed. But it was a stereo audio file (of a mono recording) they were playing back in mono and some sequences had the phase inexplicably reversed, resulting in total silence! Quickly & easily fixed. 🙂

[Spamhunter]

I believe that the way AI works, by repeated iteration + comparison, it is the only way this could be done effectively.

The only one I've tried - https://vocalremover.org/ - wasn't great, but they will be getting better all the time. Give it a shot & let us know.

[Ares Hovhannesyan]

Never heard about https://vocalremover.org/. Thanks for info. May this also can be effective  but not sure for your task Enhance Speech from Adobe | Free AI filter for cleaning up spoken audio

[Spamhunter]

That looks good, thanks.

All it needs now is for Adobe to offer it as a plugin.