No secret switches; no Druidic incantations; no Digital Wizardy?

Why do they call the company 'MAGIX'?

Pity, eh! 😔

@rickduley

You may want to try Ultimate vocal remover;

https://ultimatevocalremover.com/

This free open-source AI software is truly amazing! Its typical use is to extract the lead vocal in a song leaving the background singers and instruments so it can be used as a karaoke track. I personally use it to get rid of background wind or fan noise. The results are far better than my attempt using audio tools to suppress the noise (I'm not an expert). So, it may actually work on the audio of your movie even if the voice over is on the same track.

But since this is one of your favorite movies, why not just purchase the DVD or Bluray on Amazon and enjoy the best quality! 😀

Pensioner 😕

@Can-Dive

I have a paid for audio removal tool and there is a real chance it would remover the dialogue from the movie as well, especially if the movie has a mono sound track. You could of course edit that track to put the dialogue back in after it is extracted to a new sound track. Sounds like a lot of work.

That plugin looks cool though. I may try it to see how many instrument tracks it can split an audio file into and how much degradation happens through removing a section.

@rickduley

No secret switches; no Druidic incantations; no Digital Wizardry?

Why do they call the company 'MAGIX'?

German sense of humour? 😉😂

@rickduley

Hi Rick,

I presume that the unwanted dialogue is mixed in with the dialogue of the film, correct?

In that case, you would need a tool to separate out the various voices and eliminate the one or ones that you don't want. Spectralayers Pro 11 does a pretty good job of this. It can separate out voices, then separate out various voices to separate tracks. I have yet to try this last part but will soon as I have a video clip with someone talking in the background at one point during a speech, and I want to remove this unwanted talking.

Unfortunately, it's expensive. I got it with an upgrade to Samplitude Pro X8 Suite. That blew my retirement budget for the year.

This doesn't help you much, but at least it's a possibility if nothing else works.

John CB

@Gid @Can-Dive @CubeAce @browj2

That 'SpectraLayers' thingy looks very impressive but, as browj2 puts it, "... it's expensive."

I guess I'll just have to wait until they write the book! 😅

@rickduley @Gid @CubeAce @CubeAce

Ultimate Vocal Remover is free (UVR). @rickduley the program may struggle with your system specs but give it a go.

The program is very straight forward to use. Here is a forum where you can read a number of posts to become familiar with the process.

https://github.com/Anjok07/ultimatevocalremovergui/discussions

Basically, you load your audio file and select a specific AI model. The model will generate wav files with the extracted vocals, background, instruments etc. You may need to run a number of models to clean the audio.

I use a 2 or 4 step process to extract the vocal.

Two Steps

1. Take your song (vocals, instrumentals, all of it, the original).
2. First use Kim Vocal 2.
3. Once that is complete use UVR De-echo-normal

Four Steps

1. Take your song (vocals, instrumentals, all of it, the original).
2. Run it through UVR5 with the (MDX) Kim Vocals 2 model.
3. Take the vocal output and run THAT through the (VR) de-reverb model.
4. Take the output from that and run it through the (VR) aggressive de-echo model.
5. Take the output from that and run it through a (VR) karaoke model.

@rickduley, essentially you will be using the background wav file generated rather than the vocal wav file and possibly only conducting one step, by running Kim Vocal 2, may be good enough.

@browj2 It would be interesting to get your opinion on how UVR compares to SpectraLayers. I suspect an expert audio engineer would obtain better results than the AI models but in terms of simplicity and time commitment, UVR produces exceptional results.

@Can-Dive

This audio extraction can't be done in any form by a recording engineer by normal means.

It is only in the recent past (Last few years) that AI training from the basic audio noise cancelling applications used for headphones has been possible with a lot of additional AI training of timbrel. phase, positioning within the stereo or mono sound stage and reverb detection has allowed such AI models to subtract and isolate such signals. Passing such 'cleaned' takes can only subtract more information which due to the processes involved will only make the file being processed more mechanical sounding in nature as the timbrel information that makes up the original sound is further lost. More recent plugins than the one I have do a better job as the AI models at their disposal within the plugins are much larger. How much better I have no idea but even Spectral Layers demos I have heard so far are far from perfect.

Is the free one you are suggesting free standing or does it require a daw to work within? I am not sure how much more resources such a plugin would need to do a whole feature film compared to a five minute music track. One would have to try it to find out. Will it work within VPX or Stand-alone or will an additional daw be required?

My plugin by comparison seems quite simple.

The extract shown in use at the end will be very short due to the nature of the file.

Ray.

Hi @CubeAce,

 I am not sure how much more resources such a plugin would need to do a whole feature film compared to a five minute music track.

That is very true. It might take a while for a full length movie and you might have to chop the audio into smaller chunks. Like I indicated in my initial response, the best option would be to purchase the DVD. The audio files I work with wouldn't be more than 3-5 minutes long.

As for the program itself, its stand alone.

1. Select the input file
2. Select the output directory.
3. Select to produce a wav, flac or mp3.
4. Select the model.
5. Select if you want vocal only, instruments only or a 30 sec sample. If you don't select, it will generate all the export files.
6. Hit the Start Processing button

Thats essentially it. More time is spent reading the posts on the forum and testing the different models to determine what they actually do, the sequence to run them in and to determine if some are better than others.

Give it a go and let me know how it compares to your Mixcraft. Yours is already more efficient because the export is already loaded into the DAW's timeline. But for what I need, UVR is good enough.

BTW, is that stem split function part of Mixcraft or is it a separate plugin you purchased? If a plugin, does it run on any Magix audio products?

@Can-Dive

The plugin comes as part of the Pro version of the program. It is not stand alone and integrated into the program. It is not a separate item that comes up in the effects list and can't be used outside of Mixcraft.

I could however put a video into the daw and process it from there. Cut into chunks if that was what was needed.

I have a three day shoot coming up tomorrow so I won't be able to play much for the next week or so as I'll be busy editing.

@rickduley

Hi Rick,

If you export about a minute of the audio to a wave file, I can try it out in Spectralayers. Try to find a part that has voice-over, background audio/music and the normal dialogue. They don't have to be all at the same time. Upload to a file sharing service or I can send you a link to upload into my Magix Hub.

John CB

@CubeAce, thanks for the debrief on Mixcraft. I'm not much of an audiophile. With the help of @browj2's tutorials, I'm slowly coming to grips with Music Maker. 🤓