Computer based live rig / In search of insights

Hi,

I

Before you think about choosing a computer, think about what sort of trouble this concept can get you into. Obviously there's the chance that the computer will crash, causing a delay in your show. It can happen with any computer.

More important, though, is that with all the processing that you're talking about, you're going to be dealing with considerable delay between input and output. This is no big deal when you're mixing recorded tracks because you can compensate for it by delaying other tracks (or advancing the processed tracks) - some DAWs now do this automatically so real newcomers are never even aware that there's a problem. Plug-in latency tends to be less dependent on the computer hardware (unless it's simply inadequate to do the job at all) and more dependent on the acutal plug-in, the audio interface (you'll need a multi-channel one of course), and its driver. You can test things out with the computer that you're using now and see if you can play with everything you need running.

There's a good reason why people use hardware multi-purpose processors in live performance. Most of these have shunk down to 1 rack space today, so if you can find one appropriate for each band member (maybe the bass and drums can share the two halves of a dual processor) and an MP3 player or pocket recorder for your backing tracks, you'll be well ahead in the reliability game and a case for the processors won't be much bigger than a laptop comptuer, power supply, interface, and cables. If you have a really fast paced show and want to get your computer involved, you could safely put your backing tracks (mixed to stereo) on there and play them using a DAW that is capable of sending MIDI program change messages which you can use to set up the processors for each song.

Alternatively, you could get one of the big Digidesign live consoles that's essentially highly tweaked dedicated hardware running a version of ProTools, with a control surface. That's how they get acceptable plug-in performance in a live application. It's possible that a dedicated hardware plug-in host like the Muse Receptor. But that's bigger than a laptop computer and a mixer, and costs lots of money.

Mike makes some excellent points and that would do well to consider. However, let me present the other side...

I've been using a computer as part of my stage act for around 8 years, both Mac and Windows. I have never had a crash onstage (knock on wood!!!!!!) because I strip the computer down to its bare essentials - operating system, whatever is needed for my act, and that's it.

Latency remains an issue, but with a decent dual core computer you can get latencies well under 10 ms, which isn't really noticeable in a live performance situation, especially in terms of using effects.

The biggest problem with reliability AFAIC is the physical computer, with flimsy little jacks and a design that was intended for salespeople on the road to run spreadsheets, not work in a live performance context. Panasonic's ToughBook is extremely durable but pricey - I think around $3K or maybe more.

What you need to do with the computer is treat it like glass, and have everything connect to it via breakout boxes. For example, don't use the internal audio (you don't want to anyway). Use an external interface that connects to the computer via USB or Firewire, and secure the cables.

Looking over what you want to do, I'd advise replacing the convolution reverb with a "synthesis"-type reverb. Convolution reverbs take a lot of CPU power, and you want to minimize CPU power so you can minimize latency.

With Windows, there are several tweaks you MUST do. The first is to disable any internal wireless card, which can wreak havoc with a system. With Vista, you want to use the same setting for maximum and minimum CPU power to avoid switching the CPU between two states, which can cause spikes. Disable all unused hardware (e.g., web came) and turn off indexing, automatic updates, system restore...anything that causes the computer to go elsewhere and do a task other than live performance.

Good luck! And get a backup computer

Last time I did anything like this I used a ProTools TDM system in a Mac tower. That effectively means there is no latency at all.

Honestly you could do what you're talking about with an ancient Mac G4 and a Mix Plus system with one breakout box. It's cheap because it's older technology, it's reliable because it's older technology, and there's no latency because the Digi TDM systems have dedicated audio DSP. I bet you could set up a latency free system from scratch that way for less than a grand.

Of course you could also spend ten or twenty times that and go with new Digi stuff, too.

I second- hell, quadruple- Craig's advice to strip whaveter computer you go with down to nothing except what you need.

Even with in-line effects, live? I suppose you can choose your effects judiciously, but that means you may not be able to use on stage what you're using in the studio. Of course that would be case, too, if you switched to hardware effects for live work.

And there is no such thing as "no latency at all." The 10 ms that Craig mentioned is workable, but you don't want much more, and Sonic isn't talking about just one or two effects, but a whole band's worth.

Yeah, but you're Craig Anderton

Agree that hard crashes aren't a big risk, but haywire and poor connections certainly is. It takes some thought and organization to come up with a system that won't fall apart in your hands as you're trying to set it up at the gig. Firewire and USB connectors aren't very robust, power supplies die when you're least expecting it (often just a break in the wire where it attached to the connector at the computer end). One can easily hit the deck if someone trips over a cable - another job for the system engineer.

I've seen too many (one is too many but I've seen more than one) people come on stage with a computer and spend too much time getting music started.

Hey Mike,

Technically there is latency in a TDM system. In practice you can't notice it until you start making long, complex signal chains with heavy processing.

The latency in TDM systems is usually measured in samples and not milliseconds, if that helps explain what I'm getting at. It's a big, big difference, even with how far native has come.

Inspired from Keyboard Magazine articles by Mike McKnight, David Polich and others, my live gigging system is based around MOTU Digital Performer 6.02 running on a MackBook Pro MB133LL/A model upped to 4Gb RAM. Prerecorded audio tracks run from a Glyph PortaGig 800 7200 RPM 320Gb Pocket Drive via firewire 800. There is a Hercules 16/12 firewire 400 interface used for audio and a MOTU Midi Timepiece AV USB for distributing midi control and synth midi backing tracks.

To alleviate processor strain, vocal and guitar processing is done with external units which receive midi patch change messages at the beginning and during songs. Currently I use a Digitech GSP1101 for guitar and Antares AVP-1 and Lexicon MX400 for vocals.

There are three synths/keyboards (Roland Fantom XR, Roland XP-30 and Korg TR88) which I either play live or in addition to midi backing tracks.

There's also a midi track for the lighting controller.

It took me a while to get up to speed and put this system together because I switched to it from PC but it has proven to be a solid and reliable system. One key thing is that the Mac is not picky about when, where and how things are plugged into ports the way PCs are. I remember a few frustrating times while trying to use a PC live if I forgot to turn on the firewire interface before plugging it into the port or if I didn't plug the right USB device into the right port I'd have to reboot etc..

That might not be such a big deal normally but in the heat of a gig setup with a million other things on my mind it became annoying quickly!

In practice, you can deal with quite a bit of latency. The "speaker is 10 feet away from you so you're always hearing 10 ms of latency" principle actually works - most of the time. And it's true that the DSP horsepower behind a TDM system is pretty good, greatly reducing processing time. And since the driver always knows what it's going to be talking to on both ends, it can be optimized. This is why most pro studios have been able to make the transistion from tape and an analog console to ProTools. Still, there are some unavoidable delays, probably the greatest one being the A/D and D/A converters (since most people using ProTools on the professional level are not using the Digidesign converters).

However, in this case, I doubt that a TDM system will come into play. Since portability is a priortiy, it's just not practical. And I expect that there's budgetary considerations, too, that preclude it.

I have an article about latency tumbling around in my head. One thing that I've observed is that the numbers mean very little other than when they're large or small. You can set a system for, say, 64 samples (if it'll work like that - and I'm talking native applications now, not TDM) which is equivalent to about 1.5 ms. But you can bet your boots that the sound will come out much later than 1.5 ms after it went in.

I've never seen any actual throughput delay measurements on a TDM system (Digidesign hasn't given me one to play with yet) but I've measured hardware that I have laying around the house. 3 ms at 44.1 kHz is pretty typical from mic in to monitor out with just an A/D converter, DSP monitor mixer, and A/D converter. That won't hurt your playing much if at all, but it can disturb a fussy vocalist when hearing his voice in the headphones, as a result of the comb filtering created by mixing the delayed signal with the direct (through his head) sound at his eardrum. Most people toss this off with "I've never had a problem" but it's real.

As a great engineer once said: "In theory, theory and practice are the same. In practice, they're different."

However this has little to do with the issue at hand.

Hey Mike,

I don't really see how portability would be an issue. The G4's are perfect for live sound because they're built in a bullet-proof case with the handles built in. You can also buy rack ears for them and mount them right in a rack. The digi breakout boxes are one or two rack spaces; the TDM cards go right into the G4. The used gear is so cheap right now you could have a complete backup system on standbye and still come in paying less than a single new Mac laptop costs.

The TDM/native latency difference is not subtle, certainly not once you start to insert any signal processing into the chain. I don't mean to say there is no latency with TDM, but basically monitoring through my TDM system feels just like monitoring through hardware, even when I have some eq and compression up on the track I'm monitoring. I can even do weird stuff like monitor through autotune and a guitar amp sim, and still not hear any noticeable latency. Basically that's because it IS hardware. I don't want to beat a dead horse though. There is other information about it out there on the web.

Maybe Craig can give a little more info about his experience with native on stage here. I'm really curious to know if Craig is monitoring right through the native system and processing the signal, or if it's more of an effects sends/multitrack playback kind of situation.

"I

But compared to a laptop computer's portability? Uh uh. But that's for him to decide.

If that's so, maybe I should get myself a last generation ProTools rig to play with.

Nor is the price difference between (at least some) TDM and native versions of plug-ins subtle. That's another thing to consider. He may already have processing plug-ins that he wants to migrate to a stage-ready rig.

Like I say, some people never notice.

I've seen Craig do his "Slammin' Faders" show. He doesn't monitor as such, he manipulates sounds and tracks in real time as he's hearing the same thing the audience is hearing. I'm sure there's a slight lag in the control action, which would be roughly equivalent to playing with monitor latency.

Thanks Mike, you got me there.

Thanks very much for all your inputs!

Originally I had concerns exactly as Mike described above and most of people that I know were advising me to go with TDM based systems but I

Well, you're going about it in the right way. Test, test, test. Find the limit and then work out a way to stay well below that limit. Because on stage, stuff happens that you can't explain that never happens on the bench.

When I review something where it's relevant, I'll measure the latency that's measurable in a way that it can be measured, from input to output of the hardware. But given the variation in computers and software (both the host program and the particular plug-in), any measurement setup would be arbitrary and everyone else's usage would be different.

And you survived!

That's right. The 10ms or so of latency really isn't an issue for me, especially with control moves. With something like Ableton Live where triggering is quantized, a few milliseconds doesn't make much difference. Ditto with live guitar.