De-mix

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

pucho812

Well-known member
Joined
Oct 4, 2004
Messages
15,593
Location
third stone from the sun
This was interesting. There was talk of Peter jackson the film maker and his new process he developed for audio separation of tracks. Here it is in action.
Interesting tech, I think so. Will it make a better album, maybe?

 
I'll admit the results are great.

It's interesting, but it turned into a sales pitch for The Beatles' recordings in the end. And it left me starving for at least some look at the process of separating. Just playback of things already separated, if you know they had access to the four-track master wavs, leaves me a little suspect.
 
It is amazing what software can do these days. Almost as god as the human ear.

Cheers

Ian
In many ways it's much better than the human ear (brain really), e.g. with the speech masked by guitars linked above by John.

Ironically when it comes to better music or better sound the old school stuff still rules IMO.
 
I'll admit the results are great.

It's interesting, but it turned into a sales pitch for The Beatles' recordings in the end. And it left me starving for at least some look at the process of separating. Just playback of things already separated, if you know they had access to the four-track master wavs, leaves me a little suspect.
My understanding of some of the original process is they recorded 4 tracks, then bounced those down to 1 track on a new 4 track tape and then filled the remaining three tracks. So the original 4 tracks should still exist. Of course they recorded several instruments at once on each track so software would be need to separate those but at least they would be a generation earlier.

I think they used a similar process even earlier when only 2 tracks were available. They would record 2 tracks (several instruments per track) then bounce those 2 to one track of a second 2 track machine and add a third track so again first generation tapes should exist.

Cheers

Ian
 
My understanding of some of the original process is they recorded 4 tracks, then bounced those down to 1 track on a new 4 track tape and then filled the remaining three tracks. So the original 4 tracks should still exist. Of course they recorded several instruments at once on each track so software would be need to separate those but at least they would be a generation earlier.

I think they used a similar process even earlier when only 2 tracks were available. They would record 2 tracks (several instruments per track) then bounce those 2 to one track of a second 2 track machine and add a third track so again first generation tapes should exist.

Cheers

Ian
My memory of this once explained to me by Geoff emrick was in the 2 track days, it wasn’t stereo like we think of a 2 track, it was band on one track vocals on the other. Since emi pop records at the time were mono, no harm no foul. When they progressed to 4 track machine, instrumentation, dictated how things were combined. For example bass and tambourine could be combined because they are so far apart frequency wise could make eq adjustments focusing on one without affecting the other much. If you ever saw the Sargent pepper’s multi-track that was floating around, for the opening number the entire instrumental is on one track with vocals on another, some guitar overdubs on a third and the horns on a 4th. Iirc the crowd was mixed with one of the other 3 tracks. You can really study it for ages on how to assemble a track production and engineering wise. As it is said they went from one machine to the other, I suspect the track with the instrumental was the submix of the first machine down to 1 track.
 
Last edited:
Slightly OT but:

The Studer 90° Filter does an amazing job of folding early "stereo" Beatles down to mono far better than L+R mono.

There's so much lost Side information using L+R in these "hard-panned" recordings that the balance is affected.

"I+Q" (Studer 90°) mono has the same overall balance as the stereo presentation whereas the (L+R)/2 mono version has a lot of elements pushed back in the mix.

"Paperback Writer" makes a good demo.
 
The early Beatles studio recordings were also early days for multi-track, some odd tracking/stereo panning decisions.


JR
When you submix the entire band to one track of a tape machine, you are limited as to what can be done. Fwiw the original stereo releases were intended for the u.s. markets while the original mono releases were for the u.k. market.
 
My understanding of some of the original process is they recorded 4 tracks, then bounced those down to 1 track on a new 4 track tape and then filled the remaining three tracks. So the original 4 tracks should still exist. Of course they recorded several instruments at once on each track so software would be need to separate those but at least they would be a generation earlier.

I understand that, but I've also watched the following vids in the playlist. And the order is: first a Jackson vid, then a Beatles song, followed again by Peter Jackson. And that goes on and on.

I've watched a dozen or so Peter Jackson vids, skipping the songs, and it didn't add much to the first one when it comes to "how does it work".

I think they used a similar process even earlier when only 2 tracks were available. They would record 2 tracks (several instruments per track) then bounce those 2 to one track of a second 2 track machine and add a third track so again first generation tapes should exist.

It's AI. But does it require a super computer, a souped up PC, a server farm, or a ... to run it? Or is it a VST we can hope to buy some day?

That's the kind of detail that I'd like to see. And nowhere on Youtube, or the net, there's something to be found. Besides, everything you find seems to be a press release rewrite, except for some musings about the Beatles' songs.

I came across a site promising the same results about a year ago. Don't remember the name. The examples were impressive. But again, no word about the process, just that a release was eminent. That site is gone already.
 
In the first (only) video I watched he mentioned who did the AI for him...

As I mentioned people have been working on variations of this technology for decades. As the technology improves the capabilities expand. One friend makes a good living eliminating vocals from recording so singers can use the backing tracks. The earliest vocal elimination was simple L-R, now DSP makes it much more complicated and better.

JR
 
I understand that, but I've also watched the following vids in the playlist. And the order is: first a Jackson vid, then a Beatles song, followed again by Peter Jackson. And that goes on and on.

I've watched a dozen or so Peter Jackson vids, skipping the songs, and it didn't add much to the first one when it comes to "how does it work".



It's AI. But does it require a super computer, a souped up PC, a server farm, or a ... to run it? Or is it a VST we can hope to buy some day?

That's the kind of detail that I'd like to see. And nowhere on Youtube, or the net, there's something to be found. Besides, everything you find seems to be a press release rewrite, except for some musings about the Beatles' songs.

I came across a site promising the same results about a year ago. Don't remember the name. The examples were impressive. But again, no word about the process, just that a release was eminent. That site is gone already.
It’s a slate product 🤣😂
The most awesome thing ever because Steven said so. No more no less.
 
Back
Top