From the 18th-22nd December 2022 Trio Abstrakt and I rehearsed and recorded in competence in the Kubus ofĀ ZKM Karlsruhe. This c. 70-minute through-composed work was commissioned by Trio AbstraktĀ  and premiered by them, just prior to the recording, at the Alte Feuerwache in Cologne, on December 3rd. Programme notes in English and German are available on my website.

For the ZKM recording we had 2.5 days to get in, get organised, and rehearse. This was followed by two days of video and audio recording—not exactly bags of time but with these players and the audio-plus-camera experts of ZKM we just about managed to finish before running for the train or hitting the Autobahn (and for me, taking a plane to the UK the next morning for the first Christmas with my oldest sister since the 70s).

Benjamin Miller, ZKM’s Tonmeister, recorded the whole thing while I kept my ears and eyes on the music and score in my role as producer. Ben edited the whole project in the first months of 2023, taking care to keep the audio and video files in sync so that the video edit was possible. I did the immersive mix (see below) in the Spring/Summer. The release is not out yet but here’s a teaser:


As can be seen in some of the photos below, not only were many microphones used as spots on particularly the percussion instruments, but a 5.1 microphone array was used to capture the room also. On top of this we had the luxury of using both Sennheiser Ambeo and Soundfield Ambisonics microphones. The idea was not just to capture the instruments in the room with these, but also the sound files which are an integral part of this piece throughout almost the whole duration. (The AmbeoĀ had all of the crisp precision we know fromĀ SennheiserĀ mics whereas theĀ SoundfieldĀ was warmer but with a significantly higher noise floor. I’ve had several experiences of different generations of the SoundfieldĀ mics and, to my disappointment, I always find I move their signal lower in the mix (or even worse: apply denoising algorithms) because of the noise they bring with them. Shame.)

Triggered by Marlies (the pianist) at 63 points specified in the score, the sound files were played back through the 47.4 Meyer Sound speaker array in the Kubus. (Despite what some online sources tell you, the number of speakers in the Kubus has changed over the years and we did indeed use 47 main speakers with a submix for the subwoofers.) We thus had a mix of sound files and instruments sounding in the room and this formed the basis of the final digital mix, with of course the spot mics and the dry sound files playing a role too once it came to post-production.

This setup involved the highest channel count of anything I’ve ever done in MaxMSP to date. True, the software only had to trigger sound files, i.e. there was no live sound processing, but these were prepared in 5th Order Ambisonics format in order to provide a detailed sound location image on the 47 main loudspeakers of the Kubus system. So as the 36-channel sound files were triggered they were recorded ‘dry, as-is’ as well as in decoded form, feeding the 47 loudspeakers, i.e. simultaneously recorded through the room microphones. All in all then MaxMSP had to send 83 channels over MADI to the recording system. And it all ran like clockwork (hah, hah).

immersive audio

Ambisonics are of course a form of immersive audio but these days most people (and most marketing strategies) focus on Dolby Atmos when speaking of immersive audio. The two formats offer completely different approaches. It can be argued that Atmos is the more modern format because, besides a fixed-channel-count audio bed, it includes audio objects that can be spatialised in real-time, using the full resolution that any number of speakers in any given concert hall can offer. Ambisonics’ resolution, on the other hand, is dependent on the order of the signal, and once encoded it’s not possible to decode at higher resolutions, even though the number and placement of loudspeakers is flexible, more so than Atmos even, I think it’s fair to say. An analogy is pixel-based graphics formats like PNG vs. vector-based graphics like SVG: the latter will actually look better if printed on a high-resolution printer or displayed on a pixel-rich monitor whereas, in the absence of fancy interpolation algorithms, PNGs cannot be made to look better with higher-resolution technology: you’ve only ever got the amount of pixels (i.e. resolution) that you originally saved.

spatial audio designer

Though the sound files for performance were mixed in 5th-order Ambisonics I wanted to put Tom Ammermann’s Spatial Audio Designer (SAD) plugins through their paces for this project. In 2022 we decided on Tom’s hardware to control our new 29.1 loudspeaker system in the Neue AulaĀ at the Folkwang University here in Essen. Some of our students are comparing and contrasting the different approaches we have on offer and I too wanted to take advantage of this opportunity to mix with the SAD plugins and generate from one SAD capture file anything from stereo all the way up to 60.1 sound files, for presentation in any given playback environment, including binaural over headphones.

The process for this project necessitated translating between Ambisonics and Atmos formats such as 7.1.6, particularly when using the lush immersive reverb effects provided by Stratus 3D. I have to be honest, and this is a reflection on immersive formats in general rather than SAD plugins in particular: it wasn’t all sweetness and light. The variation in different technologies’ translation/specification of immersive audio formats, channel counts, and channel orders, led to a lot of head scratching, particularly as Stratus 3D determines its format options based on the channel count of any given audio track it is placed upon. Sadly this didn’t always tally with the format of the SAD bus. I’m not sure whether other, more immersive-audio-focused DAWs (e.g. Cubase 12) might make this easier (or even harder perhaps) but immersive/Atmos mixing in Reaper—where you have extremely flexible channel and routing capabilities but an agnostic approach to what format these might be being used for—wasn’t as easy as as Ambisonics mixing. Perhaps a 70-minute, 81-track project (plus multi-camera video) wasn’t the right decision after all for cutting your teeth on Atmos and SAD. Who knew?

more details please

Apart from the programme notes in English and German I’ve written about the image-to-sound synthesis software I developed particularly for this project elsewhere. More on the bespoke algorithms is also available in a separate post.

photos from the ZKM recording session

I can’t thank the ZKM team enough for their professionalism, expertise, and patience, including Ludger BrĆ¼mmer and Dominik Kautz, who coordinated the pre-production. It goes without saying that I was also very happy to work with Trio Abstrakt, both on the concert premiere and the recording. They went way above and beyond what you might reasonably expect from musicians young, old, professional, or otherwise.

The complete team, left to right: Benjamin Miller (recording, audio editing), Michael Edwards (composer, mixing), Salim Javaid (saxophones), Hans Gass (lighting) Shiau-Shiuan Hung (percussion), Andy Koch (camera, editing), Marlies Debacker (piano, sound file triggering), Andy Koch (camera, editing), Max Clausen (camera)


getting things into position


the 5.1 mic array, amongst other things, including Ambeo and Soundfield mics in the centre


a bird’s eye view from the Regie


view from the pianos up to the Regie


in the Regie, in the midst of discussions


all done: leaving ZKM, with moody December weather


Share Button