Sgt. Peppers 50th Anniversary Celebrations - including discussions about the 2017 remixed album | Page 2 | Fab Forum | The Beatles Bible

Take a look at this fascinating article on how they ‘demixed’ the Hollywood Bowl tapes to isolate each instrument:

https://www.wired.com/2017/03/…..ade-great/

I’d bet they’ve used the same tech on the forthcoming Pepper remix, separating all the performances that were laid down together or combined in reduction mixes. And if this one does well maybe there’ll be more remixed albums to follow. It’s the only way they can really milk serious money from the back catalogue, so I’m sure they won’t stop with Sgt Pepper .

Dark Overlord said
Isolating each instrument with their first 2 albums would be a terrible idea because they were recorded on 2 track tape. In other words, the stereo version of these albums are essentially the original multitracks and the only way to further isolate would be by messing around with the EQ or in the case of a couple of those tracks, following this method:

1. Find the original 2 track tape before the overdub happened

2. Find the overdub tape

3. Combine them and make a 3 track tape

Did you read that wired.com article I posted? It has huge implications for possible future remixes of the two-track recordings. Here:

Technological wizardry helped uncover the hidden sonics. But don’t think you can just run out and buy the same software to make your crappy Can bootlegs listenable. There’s no checkbox in ProTools to reverse-engineer a lousy recording. To get a sense of what the team at Abbey Road Studios did, imagine deconstructing a smoothie so you’re left with the strawberries, bananas, and ice in their original forms, just so you can blend them again from scratch.

To do that, James Clarke, a systems analyst at Abbey Road Studios, developed a “demixing” process to separate each instrument and vocal track from the cacophony. He isolated everything Ringo, Harrison, Paul McCartney , and John Lennon played and sang, separated it from the din of the crowd, and effectively created clean tracks to remaster. Fittingly, Clarke’s audio-modeling process used spectrograms—imagery you might associate with ghost-hunting—to bring the spirit of these live performances back to life.

“It doesn’t exist as a software program that is easy to use,” Clarke says. “It’s a lot of Matlab, more like a research tool. There’s no graphical front end where you can just load a piece of audio up, paint a track, and extract the audio. I write manual scripts, which I then put into the engine to process.”

Before tackling the project, Martin told Clarke to take a crack at a track Martin thought might give the engineer fits. “I challenged him with ‘And Your Bird Can Sing ’ on an acoustic guitar, and I knew just by being a mean, mean b*****d that separating acoustic guitar and vocals was going to be the biggest challenge for him,” Martin says. “You have a lot of frequency crossover and distortion of signal path that goes on.”

Clarke passed that test. Then came the real challenge: Working with those three-track source tapes from the Hollywood Bowl to create digital models of each instrument, the vocals, and the enraptured crowd. From there, engineers could tweak each track to create the final mix.

Separating the kick drum and bass guitar proved relatively easy, because low frequencies don’t suffer from crossover with crowd noise. But vocals, guitars, snare drums, and cymbals share the same sonic real estate with the banshee wail of the fans. The Beatles’ virtuosity and consistency helped here. The modeling process involves using samples of each instrument to help the software determine what to look for and pull out into its own track. If the recording didn’t have a clean enough version of the track Clarke wanted to isolate, he used session recordings to build those audio fingerprints. “I went back to the studio versions to build the models,” he says. “They’re not as accurate, as there are usually temporal and tuning changes between playing in the studio and playing live, but the Beatles were pretty spot-on between studio and live versions.”

After creating spectrogram models of each instrument, he loaded the files into what he calls his “little controller program.” A few hours later, it gave him a clean track of the instrument he modeled. All of those tracks went to the mixing engineer.

The above is about the Hollywood Bowl recordings, which were on three-track. Of them George Martin said this:

You would [normally] record the band in stereo on two tracks and keep the voice separate on the third, so that you could bring it up or down in the mix. But at the Hollywood Bowl they didn’t use three-track in quite the right way. I didn’t have too much say in things because I was a foreigner, but they did some very bizarre mixing. In 1977, when I was asked to make an album from the tapes, I found guitars and voices mixed on the same track. And the recording seemed to concentrate more on the wild screaming of 18,700 kids than on the Beatles on stage.

Last year, they separated all the voices and instruments from those three-track tapes, and the crowd noise too, which allowed them to remix it all over again. They unblended the smoothie.

It should be even more straightforward with the two-track EMI tapes, because they’ll be able to take “audio fingerprint” samples from the same sessions while the band was warming up.

They could not only demix the two-track 1962/3 EMI recordings, they could also tackle things like the Kinfauns tapes, the early Anthology recordings, and the Lennon demos that were used for Free As A Bird and Real Love . It’s far, far more advanced than anything they had in the 1990s, and I think it’s hugely exciting.