In the line of preternaturally talented producers to emerge from the UK in the last decade, Sam Shepherd is the preeminent renaissance man. You almost can't read an article about his work as Floating Points without feeling humbled by his many strengths. Much-loved DJ and crate-excavating record collector? Check. Neuroscience PhD? Check. Back catalogue of inescapable club tracks? Check. 16-piece band leader? Check. Command of jazz, Baroque and French impressionist harmony? Check. Well-drilled choir boy? Check. We'll spare you the full list.

With credentials like these, quality music feels inevitable. But Shepherd has the nous to stop his output from becoming a staid procession of perfection. Perhaps even more so than 2015's Elaenia, his new album, Crush, will catch Flo-Po sceptics off guard, trading the former's ECM-school tapestries for dextrous, dry-as-a-bone rhythms and an off-the-cuff immediacy that's charmingly scrappy by Shepherd's standards (disclaimer: by almost anyone else's, it's impeccably executed). While his harmonic sensibility and velveteen textures remain intact, many of Crush's best moments are coarse, off-kilter and aggressive.

Recorded in a short period of time, Crush is an outgrowth of Shepherd's solo live sets, which are centred around a Buchla system that he's designed into a massively over-spec'd drum machine. When we visited his studio back in August, he took us through the particulars of that system and explained how it fed into the creation of Crush along with other crucial components feeding into the record, such as the magisterial Yamaha CS70. But hopefully the interview speaks largely to the impact his polyglot musical education has had on his output. It should also offer approaches for the music theory agnostic to harness some of that under-the-hood understanding for themselves. As Shepherd himself is careful to point out, many of his favourite musicians have none of the skills with which he is indelibly branded. There is, in fact, hope for the rest of us.

It's a bit funny asking this given we're surrounded by all this impressive equipment but I wanted to begin by talking about harmony. The album's first track with that ninth chord and the suspension then the way it modulates to B minor very subtly, that's not something the average dance music producer is doing. So how do you go about building those ideas? Are you one of those people who can write without an instrument in front of them or does it stem from recording improvisations?

A lot of what I do is quite pianistic in that I'm initially generating ideas at the piano or some sort of keyboard instrument, then I'll write the score out for an ensemble. Often my friends are kicking about to record the parts. Most of the time you write some music and you add strings to it almost by default. That's great and I'm well into that, I've done it a lot on this album. But oboe, clarinet, bassoon, that kind of world gets a bit under-utilised. "Falaise," the opening track on Crush, is two violins, viola, cello, flute, clarinet, bass clarinet and French horn. The idea for that tune was to treat acoustic instruments like oscillators from a synth.

You sort of, shall I say, "quote" the chord with that cascading figure where each note is played in sequence by a different player.

I'm not sure how I feel about that bit now. I recorded it and thought, "Yeah, cool, it's done." Then I listened back to it and wished I'd written it slightly differently. Thing is, you have to draw a line somewhere when you're writing. I like the old idea that albums are a snapshot of where you are in time but that also means I'm already kind of over it. That bit you mention, now I think, "Whoa, damn, I'm not going to do that again. I'll do it different next time." That realisation was fairly rapid after recording it.

I think it works well as an opener though, using this declarative tonal statement.

I wanted to start with something that was very organic but then quite quickly destroy it by processing it.

You chopped it up with a gate of some sort.

It's the most basic Buchla patch. There's an envelope generator opening and closing lowpass gates. Changing the attack time of the envelope makes it feel like sounds are happening backwards. I did a pass changing the envelope shapes in real-time while the musicians were playing.

The live instruments were being sent into the Buchla as they were being played?

Yep. Usually I'd have eight oscillators going into those lowpass gates. So instead of using boring oscillators I thought, "Let's use a clarinet." But to properly answer your first question, I'll start playing on the piano to develop an idea but then generally I'll sit down and write the score on paper. I usually write to paper because it's faster to get ideas down quickly. Then at some point I'll transfer it over to Sibelius.

For those who don't know, Sibelius is score writing software. You've also just flashed me back to bad music school experiences.

Don't worry, I dislike it greatly, too. Having said that, I want to go back to Sibelius version one.

That is properly Stone Age.

I wonder if there's a way to do it without having to download Mac OS 8 or something. Version one was all I needed because I don't use the playback function. I assume the manufacturer Avid focuses a lot on recreating truthfully what a score will sound like in reality. But the processing power it requires would probably send my computer into nuclear meltdown. Who on earth has that much computer capacity?

Sibelius is useful in that it's indicative of when you've got the right notes but don't for a second think it's going to be anything like the real-life sound of musicians. What you write on the score can be rudimentary but then you give it to your friends who play instruments and it's always better. It's the best feeling in making music I think, when you give a score to instrumentalists. You might need to make a couple of changes but generally musicians are very happy to help. They'll invariably point out things like certain double stops that aren't physically possible. The violin player always has something to say.

This is the type of self-analytical question that can be hard to answer but could you identify some of the ingredients that contributed to your current harmonic taste?

I played the piano since I was a kid but by my friends' standards I'm pretty rubbish. I appreciate that's a little disingenuous. I'm OK, I can get by, but I would never offer my services to anyone as a pianist. I feel like I've got an understanding of harmony in an academic sense and I did a lot of that sort stuff growing up, looking at Bach Chorales and things like that, Baroque harmony. I got quite heavily into jazz and French impressionist stuff like Messiaen, the colours of French music of the early 20th century. I guess I'm still stuck with all that. I would like to take piano lessons again because I feel like I learnt so much about harmony through playing the repertoire when I was 18 or whatever. So broadly I would say it mainly comes from playing the piano. Also, listening to lots of records is really important. It's very difficult to pin it down.

Yeah I didn't expect as simple an answer as, "I like rootless keys because of Bill Evans" or something like that.

Bill Evans, well there's an example. There was a certain way he voiced his chords, stacking a fourth on a fourth on a fourth on a fourth then a third on top, for example, so he'd get these amazing harmonies that are open to interpretation, or "rootless" as you say. That's "So What" basically. When I was younger, I thought that was such a beautiful way of representing any kind of harmony. You can change the bass note and move things around slightly to create all sorts of things. I always found that to be a very open sound. I thought, "OK, I'm taking that," so it's in my library of chords. There are other pianists I took from of course, a lot of the late John Taylor, my piano teacher Les Chisel. Kenny Wheeler. I love these flat five chords that are super juicy. Jazz harmony in general really.

Just to be clear, I don't want it to be assumed that a technical knowledge of harmony is necessary for making harmonically interesting music.

You absolutely don't need to be able to play an instrument to explore this world. I think maybe as a kid I thought it was essential. Some of the best musicians I've met can't play an instrument. Many of my favourite musicians wouldn't consider themselves to be technically sound. But those same people can have an incredible sense of harmony.

Even though I do have that technical grounding, there's a process I use that'd perhaps be useful for people who think they struggle with chords but still want to come up with something that interests them. Basically I record lots and lots of chords completely independently of each other. The overall key doesn't matter, they can have no relation to each other whatsoever. Then you sample these chords, lay each one out on a key of a keyboard and experiment with playing them back in a different order to see if a certain combination sounds nice to you.

That's just one way of exploring new progressions that you otherwise might not have come across. For me, as someone who can get around a keyboard, I still might not come up with any of those combinations naturally. I've got those go-to voicings as you say. I end up slipping into certain progressions. But this technique can get you exploring harmonic progressions that you wouldn't have thought up, whether you know about the theory behind harmony or not.

I've done it on the new record actually. At the end of one of the recording sessions with the winds and strings I had them play loads and loads of chords. It was punishing. They were in the studio laying down hundreds of random chords. Different keys, all sorts of stuff. Then I lay them out as samples on the keyboard and hit them in a random order and end up with something pretty cool. For me personally, sometimes I reverse engineer it and rewrite it as a score and maybe use it for something else. Of course you can also use this technique and keep things really simple. Just choose two of the chords and stick a beat under it.

So there's nothing to stop anyone from exploring harmony if they can't achieve it with an instrument. Even if you play something like the flute you can't make chords, right? Obviously it's a blessing having a polyphonic instrument where you can play more than one note at a time. But yeah, sample stuff, play it back and see what you like the sound of.

Another factor is what music you listen to and how that shapes what you consider to be "normal," harmonically speaking. Say you're listening to Herbie Hancock all the time and you recognise that there's a certain chord or set of chords you like the sound of—your harmonic taste is mostly defined by what else you listen to. If you only listened to gamelan you'd think all my chords sound wrong.

What about the issue of being able to effectively channel what you want to express? I think regardless of the technical training aspect, there's perhaps still a barrier for a type of person who thinks in terms of moods but might not know what combinations of notes get them there or how to figure that out. Maybe building up relative pitch would be the best way to sort that out? Then you can reverse engineer whatever catches your ear.

I was in a choir when I was young. We rehearsed every day. They'd play a C and say, "OK, now you start singing." So you had to think, "Well, the next note is a G below that C," and you connect the dots between each note and come to know instinctively the distances between them. It gets drilled into you.

Singing is a very useful tool for anybody to get a sense of relative pitch. It's easy to take for granted when you've got it but it's massively useful. Also, I feel intuitively that everyone can sing on a basic level and build this sense up. If you can speak you can sing. It's probably the best way to develop your internal sense of melody.

Your mentioning the choir and Bach earlier made me think of that classic G major resolving to C minor move in Crush's "Requiem," which basically screams Baroque. You can practically picture the courtesans doing the Allemande.

It's funny you pick out that moment. I really like that tune because it's so simple. You could have absolutely zero technical ability and I could teach you to play it. What makes it a bit more interesting is that the Yamaha CS70 on which it's played has these sliders that move the pitches up by fifths and octaves.

Well this touches on something else I was going to ask about. One of the signatures of Crush, at least how I heard it, are these trills that almost sound like running an offset into a pitch quantiser, which makes the notes ladder up and down in this mechanical but florid sort of way. I know that's unhelpful modular synth jargon but those trills sound exactly like that. It happens on three or so tracks I think.

Well, that's these sliders on the Yamaha. I think it's a beautiful synthesiser.

The noise on it is really nice. It almost sounds like blowing, or friction on a string.

Yeah, exactly. The fader for the noise volume is so long and I only ever keep it right down at the very bottom. You turn it up a little bit and it's instantly too much. I'm thinking, "What are people doing with the rest of this fader with all this noise?!" But yes, it's one of the richest noise generators I've got in here. I remember talking to Julio Bashmore about the CS80 and he's like, "Man, the white noise on that thing!" I'm thinking, "What have we become, connoisseurs of synth noise?" But this is a CS70, I think it's a bit later than the 80. It's got a little computer in it that stores presets. It's wicked.

Back to the pitch offset things I mentioned. They work separately for the synth's left and right channels so you can have different pitches on each side of the speakers, which ends up coming off as a sort of panning-style effect. It creates melodies that sound like trills when I'm moving the sliders very fast. Harmonically speaking, if you're not getting far with lots of notes, come up with very basic two-note chords and expand them with fifths and octaves. So this piece is C-E flat, G-D, A flat-C, F-A flat, A flat-C then G-B and so on. It's very basic but adding the extra intervals with the sliders really opens up the chord.

I also think the lead line in "Requiem" is worth talking about because it expresses itself with a delicacy that doesn't necessarily come easily to synthesisers. As it develops you're controlling the timbre as if you were an instrumentalist. Obviously every synth has different parameters but could you talk about what goes into being able to pull off that sort of thing?

I think I also did that on the CS70. It sounds quite ARP-y though. For me, the ARP Odyssey has been the best one for these expressive lines. Patrick Forge turned up at my house one day with an Odyssey and asked, "Do you want this?"

Not a bad result from an unannounced visit.

This was about 12 years ago. He asked if I can look after it. That was the only synth I had back then. I made everything on it. "Vacuum Boogie," "People's Potential," all of that. I'm forever indebted to Patrick for that. It's still my favourite mono synth. There's a new, smaller one I think.

Yeah, KORG do it.

It's wicked. I love that thing. We've got three of them in the band. Sure, there's tonal differences between the new one and the old one. But all that sort of talk winds me up, nerds saying, "It's totally different, it's not the same!" But does it sound good? Yeah, so shut up.

It's also miles cheaper than an original. But to actually answer the question, it's a matter of practice and knowing the filter really well so you can control the darkness or brightness of a note based on what you want to express. There are three eras of ARP filter and I'm really into the later one, which is the least favoured of the lot from what I can understand from the opinions in the nerd world. The Moog rip-off filter was the original one where they got in trouble.

Hence the so-called "post-lawsuit" filter.

Right, yeah. They ended up with this orange- and black-faced one. That's my favourite. I got an ARP 2600 four years ago and it's basically the same thing as an Odyssey. You can patch with it and it has extra oscillators but it sounds more or less the same. The filter architecture is the same. It's a beautiful instrument to play as a performative thing.

Back to the question again—it's simply about treating electronic instruments as if they were any other instrument. They can be as expressive as their acoustic counterparts, you just need to treat them that way. So like a violinist, that means practicing with them, playing them all the time. You can't just pick up a violin and expect to play something beautiful and it's the same with a synth.

Here's a perfect example, or an example of what not to do, rather. I got all this Eurorack stuff very rapidly but I haven't really gotten into it simply because I just don't know it well enough. That's why I'm no good at it. I thought I'd be alright because I feel like I know the Buchla very well now. But it turns out Eurorack is so, so different. My friends are making amazing music with Eurorack but I'm still struggling. It was a mistake getting so many modules so quickly.

You said before the interview that you started with just three Buchla modules.

Yeah and I've only gotten one at a time since then. I think the most I ever got in one go was two. I really like the 200e series stuff, I think it's brilliant. Yes, it's digital but I don't care. The digital-analogue people can have their talk all day if they want. Are they making any music? Probably not!

Coming from a traditionally trained context, what was it like jumping over and learning the digital side of things? Or maybe you learnt both at the same time?

At my school they had three studios, EMS One, Two and Three, which is why I call this studio EMS Four. EMS One was this big fancy new digital studio that the school installed. When I say big and fancy, there was a Mackie D8b running two ADATs. Looking back, that is not good! A very bad way to work. Another studio was all MIDI stuff, ATARI computers with Cubase, a MIDI keyboard, a Roland JV1010. That was also pretty rubbish.

Then there was EMS Three. No one used it because it was in this cold corridor at the end of the school. It had a desk, an Akai S900 sampler, an ATARI, a couple of effects and an eight-track tape machine. After bugging the staff for some time they started letting me use it. I wasn't really into the MIDI stuff, it seemed like a world of rubbish piano sounds.

I thought, "Why can't I sample a real piano and use that instead? Or sample something mad?" So most of my original samples were taken from objects rather than instruments. I'd sample sounds like the fire extinguisher then pitch it down 20 octaves and suddenly you've got this thunderous sound. That was a mind-blowing experience for me.

Fast forward to getting a laptop. I had Emagic, which was an early version of what became Logic. It had the EXS24 sampler in it. Suddenly I didn't need hardware. I was accumulating tons of samples on a minidisc recorder with a little Sony mic. I knew about musique concrète and all this found sound stuff. I didn't really like the music so much but I was drawn to the idea of making music that way. I wanted to tame those sounds into something melodic. A lot of the music I was making once I was really getting going was trying to make domestic objects behave melodically. But it wasn't very good and I didn't know what I was doing.

When is this roughly?

Maybe age 13? I was really into it, maybe because it meant I didn't have to practice the piano as much. So I was making electronic music fairly early.

How different was what you were doing then to your first releases?

I was initially making stuff that was very weird, almost classical. I guess that's what it'd be classified as. Then I came to London for uni. I was going to go to music college but I got dead into science. I started going clubbing pretty quickly and I was especially into jungle and drum & bass.

Dance music has a relatively specific set of technical demands in terms of the mixdown. Did you naturally acclimatise to that, too?

I don't know if I should admit to this. There are, shall we say, certain online forums that discuss jungle that I was on where people shared productions. Basically, I made a lot of jungle when I was younger. I was sending it to everyone—Fabio, the Hospital guys. People were often very nice, saying things like, "This is great, keep going, keep progressing." But I think tacitly they were implying that it wasn't going to pop or work physically in a club. It was interesting because that community was so supportive of people learning while also being kind of militant about the music being produced well.

I wasn't doing anything right technically but I was still sending my music around constantly. I don't quite know how it happened but I DJ'd before Andy C at The End when I was 18. I sent a mix CD to them and Mr. C, who I think was running the place, phoned me up a couple days later asking if I wanted to play with Andy C on the weekend. I was like, "Shut up" and hung up the phone. I thought it was one of my flatmates trolling me. But he called back and said, "I'm not having you on, are you up for it?"

So I went down there wearing my Sunday best. I didn't know what I was doing. Fats was my MC. I was totally out of my depth playing in the main room. I'd never DJ'd out before. I get to the end of my one-hour set and I've got zero tunes left. Andy C was really slowly getting his records ready and I was stressing inside thinking, "What are you doing, I don't have anything left to play!" and he casually walked off saying, "Don't worry, play something else."

The only thing I had left was my own music. So I started playing it. It was a massive wake-up call for me. The reception was nice, people were dancing. But it definitely didn't sound anything like the other records. There was tons of reverb on it. The drums came from this Roland RD-700, which is like a concert keyboard.

This story is stressful.

It was terrible. But at the same time, it was great for me to hear my music on a system that big. Andy C eventually came up and was like, "What's this?" I said it was my own music and he was complimentary about it. So it was also a very nice experience. I knew I had to keep working to make my stuff sound right. It's a basic thing to say but hearing your music really loud helps a lot. With the production stuff, I still don't think I know what I'm doing though.

Let's suspend reality for a second and pretend that you actually don't. Does it bother you?

Kind of. I've got all this gear and I still don't think I know what, say, a compressor is really doing. But when I turn the attack time up a little bit, I know things sound kind of better. I never studied compression, I just understand the principle.

I guess doing it with a more limited hardware compressor is maybe less conducive to thinking about it scientifically as you're more inclined to change a setting and listen rather than thinking about something like the knee shape or whatnot.

Yeah, exactly. When I see this iZotope Ozone plug-in thing giving me graphs of everything that's going on, it gets overwhelming. Am I listening to music anymore? I guess it's always been about what sounds good to me in the room I'm in. At the moment, I'm very lucky to have a system in here that I know and translates well when I play stuff out. That's obviously the point you want to get to. Even when you're making music in your bedroom, getting things to a point where you at least know the flaws in the room sound is so important. This means listening to loads of records.

We actually only moved a sub in here recently and for a few days I was thought it was horrible, I hated it. But then I started listening to other people's music and realised that their stuff sounds wicked while mine sounds rubbish. Now if I make something that sounds kind of good in here then I know it'll sound good elsewhere. That of course is the whole point of monitoring, knowing your speakers.

Not to dredge up another old quote but I read that you wanted to make your studio sound like the Plastic People system. How would you even go about trying to recreate something like that?

By taking the speakers from the club. These are the old Plastic speakers.

Well that'd help a bit!

The mains here are a pair of ATC SCM150s that used to be in the club. When they shut it down I sheepishly asked if I could buy two of the speakers. I used to use these blue JBL 4343s. They're from Thin Lizzy. They have great stereo width but they're definitely a bit honky. I think at the time I was using them all my tunes sounded a bit soft in the mid-range. Now my stuff sounds too bassy because I didn't have a sub until recently.

There are certainly some very bass-heavy moments on the record.

It's because I didn't have the sub cranked up. I sent the album to David Wrench, Four Tet, Caribou—they all know their speakers really well and they all said, "Turn the bloody bass down!" And it's just because the sound in my studio didn't translate properly. But it's good now. Now I'll probably end up mixing with not enough bass.

Excessive bass was a problem with my early records for sure. Matt Colton would master it direct from tape and say, "I don't have any more tools to take away all this bass" because the dials on his EQ would only go to minus nine db or something like that. That's something to consider: mastering engineers can only do so much without dragging your track into a computer.

Another signature I got from Crush was this blurring of the line between synth and drum sound, say on a track like "Last Bloom." You'll have something resembling a kick and then an envelope will open to reveal a really characterful oscillator tone under it. I take it that must come from your Buchlas?

There was actually loads of this MAM drum machine on the record. It's not the best known machine so I was keen to get it on the album. One of my records used the KORG Volca Beats, which isn't exactly high-end, and everyone was like, "You've made the worst drum machine snare sound into the central theme of your record." But that was exactly what I wanted to do. I love the sound of that snare.

All across the record, there's this very grainy, almost down-sampled feeling drum sound.

Yeah, that's the MAM.

This was sort of like a budget version of the Vermona DRM-1 I think.

The snare drum on it is sick. I tried so hard to recreate it and I can't get close. I got it in February and immediately thought it was brilliant. My friend Kirk, he's in my band and makes amazing music himself—he told me about it. Most of the tunes are the MAM plus Buchla. The Buchla, I could talk for days about it. With Elaenia, I used it as a percussion machine. The beginning of "Silhouettes" is using the Buchla in that way.

A fairly crucial part of the Buchla system is this Roland EV-5 Expression Pedal that I've attached a battery to so I can use my foot to control a whole bunch of parameters at once. It's connected to the oscillator pitch, the decay and attack times of the envelopes, the delay time and feedback, loads of stuff. Basically it makes it sound like everything's going backwards and then it sucks back into time.

I'm very keen for someone to make a proper version of this pedal so I don't have to hack these Rolands to death. They burn out eventually. I'm playing the piano live so being able to control the Buchla with a foot means I can cover more ground. But when I was doing this record, there was a lot of hands-on tweaking of the decay times to make the sounds bloom at certain points.

On the new album, your rhythms rarely repeat the same way and there's usually a fair amount of timbral variation. Is this from using different loop lengths for each part?

The simplest but maybe most effective thing is not having the same length for every rhythm. If you've got four percussive elements, make sure they're not all the same length. It's like Messiaen's Quatuor pour la fin de Temps idea. He's got these melodic patterns of 12 notes on the clarinet but the rhythmic pattern is 13 notes long. So it goes through the 12 melodic notes but then the first note of the next melodic cycle is only the 13th rhythm. Then the second note of the melody is then the first note of the second repetition of the rhythm. Then the other instruments have their own specific rhythmic and melodic patterns, too, so nothing ever loops. It takes 200 years of them playing the tune to all land on the same beat again.

The point being that you don't need that many variables to create almost endless variation.

There you go. It always keeps it interesting and moving. With that piece, Messiaen was trying to express a sense of eternity while he was incarcerated in a prisoner of war camp. It's a beautiful story from very dire circumstances.

The fifth track on the album, "Karakul," sounds like a Buchla solo but it feels too deliberately composed to have been done in a single take. Was it built collaging bits out of different jams?

Actually the way around that was using this 225e MIDI Decoder/Preset Manager. You make your preset patches and then jump between them when you like, so you can do very specific changes. The presets themselves are stored in the actual modules so the manager just tells them which preset to address.

So the process of making this track was something like: make preset A, make preset B and so on, and then you sit back and hit the preset change button to scroll to the next preset as you see fit.

It's as simple as that really. I also absolutely love this little Yamaha Reface CS synth. Banger! The bass on it is brilliant. It's got nice detuning, too. I think it's one of the best synths in here. This and the ARP Odyssey. If I had to save two in a fire.

But you could buy 7000 Refaces for one Buchla system.

Actually on second thought that's much more sensible. Despite what I just said, I'm very aware that these things are prohibitively expensive. At the same time, people always say Buchla is too costly, then I go to their house and they have a wall of Eurorack modules. Buchlas are so excellent, so brilliantly designed, you'd need ten Eurorack modules to do some of the things a single Buchla module can do.

The stereotypical modular patch that you see from beginners—and often "experts," too, for that matter—is often a very blunt object. It's lacking the combinations of tonal and volume-based variety that approach something like an acoustic instrument. Are there things you tend to do when you're patching that contribute to creating a more complex expressivity?

I have so much respect for these modular synth people like Todd Barton, Suzanne Ciani and Robert Aiki Aubrey Lowe that manage to make them sing like a violin.

Rashad Becker has a bang-on impression of extended saxophone technique.

Right, that's it. People like this, they've mastered that instrument. You can't just pick up a violin either obviously. Years ago I bought a flute thinking, "I'll just learn the flute, how difficult can it be?" What an idiot. What kind of arrogance was that. I thought it'd take me a couple of weeks. Impossible! You can't just pick this stuff up and think you're going to get anywhere.

The Buchla is designed to make chaotic music, it's meant to make anything possible, so I'm using it in a very restrictive way. Since I'm triggering it with MIDI, those signals also come with velocity data, which helps give a patch dynamics and makes things pop out. At the end of the day, I'm making very "normal" music with it that the Buchla community might turn their nose up at. But the process for me is to try and negotiate with it, to try and calm it down. To put a leash on the instrument. For me, getting something musical out of it is about containing its ferocity.