Radiohead’s Idioteque was released in October 2000 and of their better know material is probably one of the clearest examples of what they were trying to achieve sonically with Kid A.

I’m a little bit of a johnny-come-lately when it comes to Radiohead; I was never that bothered by any of their earlier guitar-orientated rock output (which on reflection I do enjoy to an extent) and when all of my friends were creaming over OK Computer, I have to confess I didn’t really get it. Even Kid A passed me by a little.

It wasn’t until 2001’s Amnesiac that I finally began to get it. I spent the next few years going back and almost rediscovering what I’d missed first time around. Idioteque was one of those songs I knew so well but began to properly appreciate what made it beautiful.

Around the time this came out many would have probably seen it as Radiohead’s closest venture in to electronic music, and while the drums are built from sequenced modular synthesisers and the main chords are made with complex frequency modulated sounds, the composition is anything but typical cut and paste dance music, with the loop lengths deceptively complex.

Perhaps this is what constitutes good songwriting – the ability to do something which on the surface is self indulgent and make it something that people like. Thom Yorke, Radiohead’s frontman had this to say about the recording:

Idioteque wasn’t my idea at all; it was Jonny’s. Jonny handed me this DAT that he’d… he’d gone into our studio for the afternoon… and, um, the DAT was like 50 minutes long, and I sat there and listened to this 50 minutes. And some of it was just “what?”, but then there was this section of about 40 seconds long in the middle of it that was absolute genius, and I just cut that up and that was it…

Lyrically, Idioteque is has some bleak themes too with citing “paranoia of encroaching technology: the central image of the song is a nuclear holocaust as well as global warming.” It’s not just the content of the lyrics that’s offbeat too, Wikipedia adds “In the second chorus, a sample of Yorke’s previous lyrics are rearranged so that he seems to say “the first and the children” in 5/4, creating a grouping dissonance against the original 4/4 chorus.”

Much like my Everything in its Right Place tutorial, I want to attempt a quick play through, this time in Ableton Live. I’m not going to obsess over every sound and finish the whole track, but I want to give you an idea of how you might go about getting the sounds together and sourcing the samples. Let’s make a start.


The track starts with drums so that’s where we’ll begin. We’re working at 138 bpm. Greenwood uses his Analog Systems RS8000 modular to generate the drums for this track, utilising the sequencer, sequential switch, envelopes and trigger generator. I’m using Ableton’s Drum Rack but any drum computer or sampler with multiple outputs will do nicely.


Thom and Jonny playing Battleships. Image © Nigel Godrich.

For the kick it’s highly likely a self-oscillating RS110 Filter was used – a multi-mode filter that has a 24 dB/octave low-pass circuit. Interestingly, Analog Systems does have its own range of dedicated drum modules (and Jonny has been since seen with a plethora of MFB drum modules) but at the time of recording Kid A there’s no evidence to suggest he owned any of these at the time.

We can approximate this with Live’s Simpler. I’ve recorded a dummy audio clip and added it to Simpler in order to employ the self-oscillating filters added in Live 9.5.

After adding the silent clip to Simpler, I’ve enabled the MS2 low-pass filter taking the frequency down to 30 Hz and the resonance up to 120% and drive at 24 dB. Use the internal envelope to ping the filter – a short decay with no attack or sustain should get you close. Here’s a much more detailed video explaining how it’s done:

It’s difficult to get much closer but I’ve added some of Live’s stock plug-ins to “dirty it up” a bit more, with some Overdrive and EQ boosting around 100 Hz.

For the snare and hi-hat I’ve used instances of Analog utilising the Noise oscillator. For the snare I’ve used Filter 1 to low-pass the sound with another short, sharp envelope opening it. I added some ring modulation (perhaps akin to the RS20), EQ to brighten, Reverb with plenty of early reflection and snappy Compression.

The hi-hats are made similarly but using a resonant high-pass filter with no envelope modulation. I’ve used the linear envelope for the amp section and added some velocity modulation to simulate the volume changes. The whole drum rack has some Dynamic Tube distortion, Erosion, Saturator and Glue Compression to homogenise the sounds together.

Screen Shot 2017-12-12 at 19.17.53.jpg

I found this excellent video researching this piece – YouTuber DK goes into a lot more detail on each sound and how to program the beat using a eurorack modular:

The essence of the sound is not hugely complicated, although tough to get much closer without undue amounts of work. What I find more interesting is how the beat is generated from three rows of the sequencer (which might explain the odd loop length of three bars) and using the sequential switch to navigate between them.

Finally on the modular topic… Here’s Jonny!

Mild und Leise

Let’s move on to the haunting main theme. The chords in Idioteque are sampled from American composer Paul Lansky’s piece titled mild und leise (German for mild and quiet). The sample starts at 0.45:

There is in fact another sample from Arthur Kreiger’s Short Piece. Both mild und leise and this appeared on a compilation CD Radiohead guitarist Ed O’Brien picked up around the time of recording.

While Lansky has confirmed the piece is based around the notorious Tristan chord (from Wagner’s opera Tristan and Isolde), the portion Greenwood sampled for Idioteque does not contain it, as sometimes erroneously claimed by other articles. Lansky has also commented that the piece was based around George Perle’s 12-tone-tonality concept, whom he was a student of.

N.B The Tristan chord is a particular voicing of a half diminished chord which originally appeared as F, B, D# and G#. What’s key about it however, is the voice leading. Idioteque’s main chord sequence is all based on inversions of Eb major 7 (Eb, G, Bb and D), see below:

chords 2-1.png

Screen Shot 2017-12-12 at 17.48.38.jpg

The last chord is actually held for two bars, creating a five bar loop.

It’s likely each chord was loaded into an Akai S3000 or S3200 and triggered one by one (you can see live footage of each chord being triggered chromatically on a small MIDI keyboard).

We can use Live’s Sampler for this task: I’ve loaded in the Mp3 into Live and sliced each to its own clip, then adding said clips to a new zone in Live’s Sampler. I’ve used the amplitude envelope to shape the chords with some longer attack and release stages and transposed the samples down a semitone. I’ve added some light Glue Compression to beef the chords up a little and Live’s Utility to narrow the stereo image:

It’s possible to get a similar to the original sounds using Live’s Operator or an other FM synth. At this stage, I feel it necessary to use the same trigger warning I used when tackling Aphex Twin’s Xtal: This sound is NOT perfect, nor do I claim it to be. Please leave my poor comments section alone if you think the sound if miles off.

For Operator A, I’ve added in large amounts of the 2nd and 4th harmonic, with some 5th harmonic too. Operators B, C and D are all tuned around the 2nd harmonic with amounts of detune. Use lengthy attack stages in the amplitude envelopes to simulate the chords fading in.

I’ve added some band-pass filtering and Spread from the Pitch section (unison detune). Lastly I’ve disabled the oscillator restart for each oscillator.

The reality is the original is much thinner, and the stereo panorama is less focused, with each voice almost panned differently. There is very little literature on the synthesis capabilities of the original synth used (more on that below). This is my best guess having only a quick listen. It might be possible to get closer with the Synclavier which I will try at a later date, but I’ve tasked myself with keeping this entire tutorial to within Live’s capabilities for now.

Lansky has written about his experience with the sampling of his piece on his Princeton University blog:

mild und leise was composed in 1973 on an IBM 360/91 mainframe computer. I used the Music360 computer language written by Barry Vercoe [inventor of CSound]. This IBM mainframe was, as far as I know, the only computer on the Princeton University campus at the time. It had about one megabyte of memory, and cost hundreds of thousands of dollars (in addition to requiring a staff to run it around the clock). At that point we were actually using punch cards to communicate with the machine, and writing the output to a 1600 BPI digital tape which we then had to carry over to a lab in the basement of the engineering quadrangle in order to listen to it.

You Can read more about Frequency Modulation here.


Here’s how the drums and sample work together. I’ve added the Kreiger sample for the intro, leaving it as an unwarped audio file and tuning it down two semitones.

It’s by far perfect, as it sounds like there has been additional drum added and plenty of clever studio trickery and mixing, from the talented Godrich and co.

I’ll leave you with an amusing video of Radiohead train-wrecking Idioteque back in 2012. Who knows what went wrong here. All I’m saying is that modulars can’t be trusted.