Minor Science: The Language of Sampling

Second Language, the debut LP by Berlin-based musician Angus Finlayson, aka Minor Science, is one that focuses on recycling, development, and complex shifting of states. Sonically, it’s a welcome return to IDM, with complex, dynamic and agile beats interspersed with inventively used samples. As a former journalist, Finlayson has applied his inquisitive nature and gleeful curiosity to that of sound, and in particular their many sources. The resultant process has yielded a record formed from an endless catalogue of samples, field recordings and recycled compositions, with themes centered around concepts of differing languages and their convoluted relationships.

An overriding impression of Second Language is that it’s elaborately produced, and although certain elements are, it’s Finlayson’s time, energy, and discipline around collecting samples that are the real backbone of this record. Talking about how he made Second Language, Finlayson goes into detail about sourcing samples, using them with Sampler, and then creating the potent percussive patterns that push his music to its limits.

Where did the idea of doing an LP centred around translation come from?

When I moved to Germany five years ago, I started learning German. I got really into it, and into learning languages in general, and this has since been an ongoing feature of my life. I find the mystery of as-yet unlearnt languages, and the thrill of opening up new zones (linguistically and culturally) by learning them, pretty addictive. When I started making the album I didn’t plan for this to factor into it, but as I worked on it I began to see certain ideas about translation, and the relationship between a first and second language, reflected in what I was doing. I then started to drill a bit deeper into these ideas, and used them to help tie together the tracks I’d made into a coherent record.

How do you use sampling to convey these concepts, and what processes do you use?

Sampling is pretty fundamental to the way I make music. I have tons of folders full off self-made sample collections gathered from various sources over the years that I draw from extensively. For instance, one collection comes from a job-lot of cassettes, mostly funk, soul and R&B from the 70s-90s which I bought off eBay and spent hours sampling using an old tape deck. Others might be from certain online sources, things that are esoteric and have some meaning to me.

For example, the seed of the track “Second Language (Intro)” came from an old, out-of-print record I found a rip of, which I won’t name for copyright reasons. I first combed through it pulling out snippets which interested me, and then filed these by type. I started the track by throwing a bunch of this material into a project and seeing what stuck. In this case it was some guitar chords, and atmospheric, foley-type sounds.

I became interested in this feeling of contrast and dislocation by laying the material out in sequence – a feeling of being plunged into different ‘real-world’ spaces, each of them seeming ‘authentic’ on its own but then superseded by another, completely different one. To my ears the end result was a kind-of cobbled-together reality which didn’t correspond to the ‘real’ world these samples had been captured from. In that way there was a kind of process of translation going on, a transmission of these original sounds into a new context, where something of their ‘correctness’ was lost, but something else was gained.

The sampled material I mentioned wasn’t enough to fully execute this idea, so I then went on a hunt for more that would fall into the categories of ‘guitar’ and ‘foley’ – things in other sample collections I’d made, and things I could find online. I spent hours looking for guitar samples with different feels and fidelities and put together a new collection of guitar chords. In the final track, “Voiced and Unvoiced”, you can only hear a handful of these.

On these “Second Language” tracks, what synthesised sounds did you add to the samples, and are there additional field recordings?

The field recordings are samples, though I did add a couple of my own, including a recording of an intense thunderstorm. That recording has appeared in quite a lot of my stuff, I suppose it’s a little easter egg that nobody would ever notice if I didn’t point it out. Sometimes samples find their way into things because of the meaning they hold for me, though often it’s just that they sound good.

Opening the project for “Intro” I see that a lot more of this is samples than I’d remembered! Some instances of the melody are played on Instrument Racks I’d built out of samples from that same source. Others are synthesised, in particular from the Waldorf Blofeld which I used a ton on the album. Then there’s some synth bass and some drums (also samples, as with all my drums).

The other two “Second Language” tracks get a little harder to unpick, as they’re kind of hybrid projects that incorporate stuff from the “Intro” track plus stuff from two old projects I’d never got around to finishing. On “Tender Phonemes” the key parts are from some of the Arturia synth emulations. Otherwise, I guess a general rule of thumb with my stuff is that it’s built from samples – even some bits you wouldn’t expect.

How do you catalogue samples so that they are easily accessible and relevant to what you’re making?

Categorising samples in a way which makes sense to you is almost as important as acquiring the samples themselves. In this way, unfortunately, I’ve realised that sampling resembles many other activities which take up my time: categorising music for DJing in Rekordbox, keeping on top of my emails, organising my life through an app like Evernote. Life right now is basically an unmanageable overload of information, and all we can do is find ways of taming it and navigating it to get the outcomes we want. 

A couple of decades ago I guess sampling was more about rooting out some gem of a breakbeat from a forgotten record found in a basement. Now, I can download hundreds of great breakbeats in an hour. The question is how to navigate this superabundance creatively, and not get stuck in browsing mode when I should be making music.

At the moment my samples are arranged in five main categories: drums; my sample packs – these are collections I’ve made myself; other sample packs – those made by other people; voices; and a folder of raw sound design audio I’ve bounced out and saved for potential future use. I’m constantly tweaking this system as my own priorities change, or as new material comes into the collection which complicates the categories I’d previously worked out. I’ve found that reorganising old sample packs according to new, more intuitive logic can be just as inspiring for my music-making as acquiring new samples or sounds.

How did you process the vocal samples in “Voiced and Unvoiced”?

The voices come from a YouTube video I ripped, which explains the difference between voiced and unvoiced sounds in English. I dumped the recording into a Sampler patch I’d made which scrambled it up. The idea was to get it just on the edge of intelligibility. You can pick out the occasional (hopefully suggestive) word, but not so much that it becomes prescriptive. This scrambling process was something I’d done a while before: I spent an afternoon running a bunch of voice recordings through various Sampler patches, recording the results, and saving them in a sound design folder. This technique is something I stole from my older brother, who also makes music and has taught me a ton of stuff on the technical side.

Here’s a breakdown of the Sampler patch I used: First up, it’s important that the sample is a long passage of speech, with silence in the background and no other kinds of sounds in there. Then, one of the Sampler’s internal LFOs, with a ‘random’ curve, controls the loop start position, so that the playhead skips randomly through the file. Then the modulation oscillator is engaged, in FM mode, with a randomized LFO (this time it’s a Max for Live LFO) controlling the frequency of it. Then, to ‘randomize the rate of randomness’, if that makes sense, there’s a second random-curve Max for Live LFO controlling the rate of both the first Max for Live LFO and Sampler’s internal LFO. The aim was to generate a stream of audio which doesn’t noticeably ‘repeat itself’ whilst also keeping the randomness contained within fixed parameters, so that all of the material produced is roughly of the same ‘kind’.

How do you go about cataloging voices and vocal fragments?

I haven’t found a good answer to this yet! I’ve made various attempts to group vocal samples by certain themes or possible applications, but none of them quite work. So I currently have a bunch of odd and possibly conflicting sub-folders inside my voice samples folder: ‘acapellas'; ‘animals’ (I went through a phase of collecting barking dogs etc.); ‘word whiskers’ (wordless byproducts of speech – ums and ahs, sneezes, breaths are in here). Then there’s one folder full of individual words, each word a different sample. I had the idea of collecting enough of these that I could build sentences from them, but never got far enough with the collection. And there’s a folder of longer speech samples, of the kind used in the Sampler patch mentioned above. After that, there are a couple of folders of phrases I’ve chopped out of things which seem somehow meaningful to me – I haven’t worked out how to use these yet. 

The concept of stripping down unfinished work for parts, in addition to using similar beat melodies across several tracks over the album feels very ‘meta’. Do you think producers are too cautious in terms of reusing pre-existing content to furnish new tracks?

I actually kind of think of it as the opposite of meta. It's like, if you make furniture and there are some pieces you never finished, why leave the materials gathering dust when you could rip them out and reuse them? It's practical. To me, the 'meta' approach is becoming too self-reflexive and getting hung-up on the idea of repetition across your tracks, and letting this fear of repeating yourself prevent you from making practical creative decisions which will help you get your tracks over the finish line. I say this as someone who had this exact hang-up for a long time. Freeing myself from this neurosis and learning to reuse material in pragmatic ways has made my creative process much less tortured.

This approach also includes saving sound-generators I’ve built as Instrument Racks in the User Library, recalling them in new projects, and seeing how the new context changes them, which it often does, in surprising ways. (I think I stole this tip from M.E.S.H.) If you trust your creative instincts to guide you away from simply repeating yourself, then the parts you reuse will always end up changing and developing in the new contexts you throw them into. At least that’s what I’ve found.

I’m also very interested to find out what your approach to percussion is, let’s say for instance on “Balconies”?

“Balconies” is an interesting choice because it’s one of the more complicated projects, drum-wise. With the album I got really into making my arrangements switch, as in there’s often an ‘A’ and a ‘B’ section, with different feelings of space/intensity, which I alternate between. This involves having two different but complementary sets of drums. In the case of “Balconies” this is achieved partly through doubling the kick-drum with different things: lighter percussion in the ‘A’ sections, then heavier claps/snares in the ‘B’ sections. Then there are a ridiculous number of different hi-hats. Looking at the project now, I count 15 including rides and shakers, plus other percussion and incidentals. The aim is to keep things moving and building throughout the track whilst also keeping some sense of consistency. I spend a lot of time in the arrangement phase trying different combinations of kick-snare-hi-hats-bass-and other to see which ones sit together and contrast or complement whichever section is coming next. It’s a bit of a jigsaw puzzle, and in this case it took weeks to slot into place.

Can you talk us through the Rephlex-like drum solo of “For Want of Gelt”? Did you piece together each kick painstakingly, and how did you go out EQing each one? 

This is another instance where having tons of well-organised samples to draw on was helpful. A lot of the one-hits were thrown in from drum machine sample packs. Then it switches to live fills – these are from some breakbeat packs I've got. It was a question of going through and finding the ones which fit the energy and sonic balance of what was already there. In general I place a lot more emphasis on sound selection than on processing. I try to pick sounds which broadly fit with what’s around them. So in this case that meant that the individual hits, at least, didn't need much treatment once I’d selected them. Each sample was individually EQed, compressed, and adjusted for volume, which took a while.

I'm a bit of a glutton for punishment when it comes to this sort of thing. In the same way that I'll happily fry my brain sampling cassettes for hours on end, I find the mind-numbing labour of building a 20-second passage of audio like this really rewarding, providing the results are good! Laptop production offers a lot of potential shortcuts, many of which I take gladly, but there’s a level of second-by-second precision which you can only achieve by zooming right in and clicking around. For the fill, I could have just piled samples into a drum rack and fired them out semi-randomly, but I wouldn't have got the same accuracy of sound balance and sequencing. I think you can hear the hours of labour in the finished result. Or maybe I’m just kidding myself.

Keep up with Minor Science on SoundCloud, Facebook and Twitter.

Text and interview: Dan Cole