skip to Main Content

Sleep Token



Full Production, Mixing & Mastering catered to by George Lever here at G1 .


Working with Sleep Token is always a pleasure. I get this deep seated selfish enjoyment when we start new projects, mainly because I’m safe in the knowledge that the experience is going to be beautifully intense and incredibly rewarding.


We started working on ‘Two’ just before winter started in 2016. Working on maturing the structure and fleshing out tones. Its no surprise that when it came to the final tracking that we changed the sonic balance almost completely (this is a running theme by the looks of things)

Time restrictions led to us using programmed drums, however we took samples of the real kit to sequence so that it didn’t ‘feel’ completely robotic or lost.

Guitars last time around resulted on leaning on my Kemper, however this time we used a live amp and cab for the majority, the momentum from the tubes / movement of the cab is something I’ve had to accept is difficult to recreate ITB. Its easier just to go down the most direct route. I had recently purchased a one of a kind Randall modded amp from a friend in the states. It turns out its an early prototype from when Fortin partnered up with Randall, its not a looker but its a box of pure filth. The Fortin was fed into my Mesa OS, placed an SM57 on the cap joint of the cone and off we went, pretty straight forward really!

Bass is a hybrid setup between two real basses and then a 3rd lane of audio from a programmed bass. Because of the way the song’s have been written, keeping the sub lows consistent is vital to ensuring the song flows without feeling disjointed section to section / genre to genre.

Vocals is where the most change happened (from a tracking perspective). Originally we had finished and finalised everything with my go-to setup. Modded Oktava > Germanium Chandler > Empirical Labs Distressor. And for the most part it worked. However halfway through the process I bought a Slate VMS. A modelling microphone. Without going into too much detail, its a microphone that can imitate more expensive vintage setups and it does it without any issue whatsoever.

Obviously when testing it out and comparing, the VMS won without breaking a sweat. So back to tracking all the vocals again.

(The vocalist is amazing however, the repeat performances were just as good, if not better than the original final takes) 

Gear Used


Ibanez Prestige (7 string) – Bare Knuckle Aftermaths
Fortin Modded Randall
Diezel Einstein
Mesa Oversized Cab
Beyerdynamic m201 & 57


Slate VMS – AKG C12 Emulation into Neve Pre


Sampled the original kit, programmed performances.


Fender Jazz Customshop
Dingwall Combustion

Synths / Samples

Logic Stock
Spitfire Albion
LA Scoring Strings


Everything with Sleep Token (from a mix perspective) is orientated around the vocal. The vocal here is god and attempting to get the mix to flow any other way would have been very strange indeed.

Sleep Token were very clear and concise about the structure for the mix, Vocals, Samples, Drums then everything else equal after the fact.

Very much like how pop is built actually. It turned out to be the most happily balanced mix I’ve worked on. I expected to have a harder time balancing moving between a few different genres through out the track but in all honesty, it just worked. Which in itself is testament to the quality of the song writing.

We worked on a few different ‘perspectives’ for the mixes, however in all honesty. The mixes came together with very little effort and the tracking stage defined a lot of the movements that would happen later down the line.



Vessel doesn’t give quotes. – ST


I ended up going against my usual workflow for the master chain, trying out new things always lends to another perspective / approach / result.

The signal chain looked something like this;

Slate VMR – Neve EQ
Focusrite Red 3
Slate VMR – Custom EQ
UBK-1 (the density control on this is awesome!)
Izotope Limiter
FabFilter Limiter


If you haven’t already checked out Sleep Tokens first release ‘One’. You should. ‘One’ was also produced here at G1 but resulted in a darker sound overall!

– G