Faster Than Sound 2010
I just got back from England last night. Tod Machover had an Artist-in-Residency with Aldeburgh (pronounced Orld-bruh) Music and we worked to create a new piece called Spheres and Splinters for cellist Peter Gregson based off of Keith McMillan’s K-Bow. It was a collaboration with Ash Nehru of United Visual Artists, and involves awesome live visual components as well.
For the gig we put together a 20 channel ambisonic system which we had to move in a period of 12 hours from Britten Studio in the Hoffman Building at Snape Maltings to Hall One at Kings Place in London. The system performed admirably and was based solely off of a single mac pro. We used an additional beefy mac book pro (Tod’s) to run the analysis system, all authored in Peter Torpey’s OSC mapping software. The audio system was a complete d&b rig of E-Series speakers and a Soundcraft Vi1 desk. I learned some tricks on this trip and we did some interesting things with some cool gear, so here’s a quick summary and some take-aways!
First, the gear. Soundcraft has some really great products. The first real desk I ever bought was a 16-channel LX7. Since then it’s been a swath of desks: Live 324, GB8, K2… all really nice to use, reliable and incredible value. After using a Studer Vista 5SR for the opera, I thought it would be cool to try out the Vi Series. The Vi1 is a really really cool idea. It’s for those times when you are space challenged, but need some serious channels. It’s got all the features of its big brothers the Vi4 and Vi6, but the screen layout is a little different and there’s certain limitations. You get only 64 channels (vs 72+ for the Vi4/Vi6) and 24 buses (vs. 32 for Vi4/6). There’s also only half the effects units (4 total), possible matrixes (8), GEQs, etc..
However, considering its a quarter of the price of a Vi4 rig, you still end up with a remarkably solid system with more channels, outputs, and better sounding effects, pre-amps and converters than most other comparably priced consoles (SC48 or M7CL, for example). Best of all, the Vi1 takes Studer D21m expansion cards in its double height expansion slot. We used an optical MADI card in addition to the analog IO provided locally. This is really the optimum board for smaller ambisonic systems because its so easy to get digital IO. It also has some really interesting features that I haven’t seen in other desks:
- Tie-lines– allows one to patch computer input (digital) directly to analog output or vice versa. This is exactly what we did in Monaco on the Vista. It’s pretty cool to do it on a desk literally a 10th of the price.
- Abstracted Inserts– allows one to swap the patch of an insert (to IO connections or internal effects units) without channing individual channel assignments. Super useful.
- Vistonics– I really got addicted to this on the Vista. For more information check out Studer’s website. Essentially, the knobs are on the screen, so it’s not turning one thing and looking somewhere else. The Vi1 made some sacrifices here. The knobs are laid out horizontally because of the display ratio. Still, not bad to use and definitely possible to get used to! But, seeing parameters for a currently selected channel above other channels is a little confusing at first.
We made extensive use of the snapshotting capability, because the two events we did were an electronic music performance (with 7 different pieces and setups) and the TEDx Aldeburgh Music Conference. All went smoothly and the Vi1 has a full complement of channel scope and isolation features which saved us in a couple situations. I had the awesome chance to mix an impromptu Imogen Heap performance with this desk (and with Tom, Aldeburgh’s sound person!). It was a blast and he and I both were able to learn our way around with very few trips to the manual.
All in all, the Vi1 is a serious contender to both M7CL and Venue SC48. While it doesn’t have the number of faders of the M7 or the plugins of the Venue, I can definitely say that it sounds better than both and the UI is intuitive and efficient. Before I go on to other things, I should mention that Tom Der at Harman US was a great help in verifying the specs of the desk before confirming our rental contract in the UK. Thanks, Tom!
RME to the Rescue!
I used my first RME audio interface on this project. Having used just about everything else out there (and seeing as our SSL cards were on the boat from France), I thought it was about time. A big thanks to Chris at Parsons Audio for being able to hook us up with an HDSPe MADI Card. The short story is: the HDSPe card was rock-solid, had stable drivers and software and useful features. The onboard routing matrix is incredibly cool and pretty useful for very low-latency routing. (For example, to patch desk outputs directly to speakers in order to bypass ambisonic DSP.) Combined with the Vi Series tielines, the super-flexible routing is a really handy tool. On top of that, it’s possible to record snapshots of the routing configurations and change the physical mix assignment of each computer output.
The AutoSync clock system is useful because the card maintains its own clock regardless of the presence of an external clock. When one exists, the card locks to it, but in-case it fails, the clock keeps running on its own. This greatly simplified our power-up sequence because the computer would lock to the desk as soon as it was available and we didn’t need to wait for various components to boot up before others.
The HDSPe MADI feels like a more polished product; even the VU meters in TotalMix seemed to feel slightly more professional and responsive than almost all other platforms I’ve experienced. I’d say that’s the definition of a good experience. I suppose you get what you pay for. Still, it is more expensive than the SSL MADI Xtreme128, which has double the inputs and outputs. The Xtreme128 is a no-frills card that gets us 90% of what we need, so unless we have special need for flexible low-latency routing, I’m a bit too cost-conscious to shell out additional bucks for half the IO. If RME did a cost-competitive card with 128 IO, I’d be completely sold!
The Gigs/The Systems
There were (as mentioned above) two gigs. The first was TEDx Music Conference, hosted by Thomas Dolby and with some big names (Imogen Heap, David Toop, William Orbit, Tim Exile, etc…). This was a fairly standard affiar with lots of Keynote and playback. Actually the only one who had PowerPoint was Imogen Heap! We did a quick demo of the ambisonic system and played Michael’s dream from Skellig.
The second gig was our real show. It was part of the Faster Than Sound series, put on by the incredible Joana Seguro. The piece used seven continuous data parameters coming from the K-Bow (accelerometers, grip data, etc..) to add electronic accompaniment to live Cello. We used a Mac Mini on stage to grab the Bow data via Bluetooth. The data was sent via OSC to Peter’s mapping system which then sent more OSC to the main DSP machine, running Kontakt, MaxMSP, our custom Ambisonic DSP inside Digital Performer, and Ash’s D3 System (UVA), running the visuals!
The DSP box had a MADI connection to the desk and we used 16 mono groups to route into the ambisonics system with returns coming back on channels and routed directly to the d&b amps using post-fader direct outs. This let me put the whole surround sound system on a VCA. I could have used tie-lines, but this was useful because I could do basic EQ and delay for the surrounds on the channel strip.
I was able to save a huge amount on shipping by having Joana and Tom rent a Mac Pro and just bringing my drives. It was easy enough to reactivate most of our software licenses for the new hardware and it meant that I didn’t need to argue with the AA and BA to get them to allow my overweight bags! I will definitely use that trick in the future.
The effects for the piece were based on two concepts, delay and sample playback. Tod has incredible banks of samples, so we were able to create scripts in Kontakt to dynamically spatialize many single note samples to create swathy chords and abrupt harmonics. Using Peter’s mapping system, we were able to analyze the data and audio coming from the bow and choose clusters of samples that were similar to the pitch and timbre coming from the cello.
Then we used a classic (for us) effect created by Charles Holbrow, which controls the randomization of multitap sliding delay lines to add our own flavor of really unique spatialization, pitch shift, and delay. Charles makes really incredible stuff and his effects have many different sounds depending on how they are set to behave. We used a very slow delay time ramp at the beginning of the piece for a breathy, chorusy sound and routed the outputs of the delays into a harmonizer (Antares Harmony Engine Evo) to constrain their pitch. Later on, we did a “splintery” section where ramp times were low and so bits of live cello appear scattered, abruptly spaced and transformed very differently in many locations. The outputs of all these effects were routed into ambisonic encoders and there was some automation on the encoder position as well to keep things moving. We have started work on a really cool ambisonics automation environment. It allows procedural control of encoder and decoder locations to create moving clouds of encoder points. We’re going to play with that more later on, but it was very interesting to experiment with in a basic way.
Digital Performer handled most of the serious DSP; we used the MasterWorks collection to add some sheen to the final sound and it served as a virtual sound check environment so Peter (Gregson) could hear his cello sound while we tweaked it in surround sound. We used ProVerb’s 5.1 Convolution engine to do our reverb as well as effects inside Kontakt to easily modify our 16 channel sample stems.
Peter’s Cello presented some issues (all good ones though!) We originally intended to use his electric cello, but it just didn’t sound as good as his acoustic. The house rented us a DPA4061 mic to use for the acoustic. It was not my favorite of microphones. I’ve used an AMT S18C with great success, but this DPA just didn’t sound that great. We ended up finding sort of a magic bullet; a De-Esser! It allowed us to really control the high-end of his cello. Luckily the Vi1 uses the Studer compressors and EQ, so we were able to do all of our processing right on the channel strip! It was quite a strip, but in the end we really grew to like it. I’ve always been in the camp that thinks there is no DSP substitute for a good microphone, but we really got pretty far with this one! We are going to try and make a serious recording of the piece (which I’ll do my best to release in 3rd Order Ambisonics and stereo!) Stay tuned for some photos as well!
I learned a lot on this trip. I won’t even go into the political issues and house tech issues (or at least, I’ll save them for another post!) Here’s the four audio/systems points that I really took away:
- Don’t take machines with you! Just rent a box at your destination and take drives. It’s so much easier and lighter. But make sure to test your boot disk on the right generation of machine and check that your licenses can be re-activated on new hardware. These are two very simple things to do and they save lugging 80 pounds of extra hardware to the airport!
- Using Post-fader direct outs for Ambisonic returns is awesome and allows really good flexibility! Tielines are another option, great for sends that don’t need EQ and Delay. Bussing out 16 encoders leaves 8 extra groups on most mid-sized digital desks for Effects sends, Foldback and Matrix Outputs.
- Having trouble with high-end on a stringed instrument? Try a De-Esser. It’s really great at allowing detail in higher registers, while clamping down on that annoying 5Khz slam that comes from contact mics and transducers in lower registers.
- Using Bluetooth hardware? Make sure to locate your devices as close as possible. We had a tough situation with a dropped connection. The next night I put the mac mini under Peter (G)’s seat. It was much smoother and we got better battery-life from the Bow as well!
Here’s some awesome photos of the project by Jana Chielino!