Support |
At 8:48 AM -0400 1/22/98, Curtis Bahn wrote: >>kim flint wrote >>One thing kind of troubling about the computer based systems for >real-time >>use is the latency problems. (meaning the time it would take for audio to >>be sent in and sent back out again.) On the MSP site some typical >latencies >>are actually listed as: >> >>Audio input to audio output latency on a 9600/300: >> Using the Digidesign Audiomedia III: 46ms >> Using the Sound Manager: 294ms >> >>that's definitely in the range where you would notice it in some >>situations, especially with looping and trying to maintain precise >grooves. >>And this is on a very fast (and expensive) system! I know that PC's >suffer >>from the same problem. The audio has to go through a lot of operating >>system to get to where it's useable, and a lot more operating system to >get >>out again. >> >>So those of you using systems like these for real-time audio I/O, how do >>you deal with that? Are you able to operate it with any timing precision >>for real-time audio events? And I don't mean hard disk recording where >the >>system has opportunities to compensate for the latency. I mean audio >really >>going in and out, like you might have on a typical stand alone audio >>processor. Anyone? >> > >Latency can be a real problem sometimes. Especially on a powerbook, the >technology is not really there yet to have an inexpensive self-contained >system that can deal with all the audio processing and performance I/O in >real-time without some significant lag. >In some cases, like making rich ambient textures, the latency just desn't >matter and it is a reasonable price to pay to be in control of more >aspects >of what your signal processing device is doing. We're certainly in a >transition period between hardware and software based models for our toys. >It's pretty exciting. Especially in performance areas like looping (which >isn't that "expensive" in terms of computer processing and won't incur as >much latency as more complicated signal processing) this technology is I basically agree with you about things moving towards a more software oriented environment. Whether it will all be running on general, yet focused musical/audio devices or on general purpose computers seems rather questionable for quite a while, if ever. You say that something like Looping doesn't require a lot of processing power, and having developed them, I'd say that's not really correct. It mostly doesn't take a lot of signal processing power, that's true, but it does take a lot of processing power to maintain the real time control. At it's core, Looping is not so much about dsp and signal processing as it is about control over data and functional execution. It requires a true real-time operating system to ensure that any possible event requested by the user can be executed at any point in time, within a guaranteed amount of time. Such an OS expends a huge amount of processing cycles making sure that this is possible. Devices like the echoplex, jamman, and boomerang use all of their processing power to make sure that when you tell it to do something, it gets done within a very short amount of time. On the Echoplex, for example, this maximum latency is only 1.5ms. Windows and the MacOS are not real-time operating systems. They are genearalized to do a lot of different types of tasks. They have some real-time services available, but not much. And those are not as good as something designed specifically for that. That might change some, but not much in the near future. Most of the services that get added have to do with real-time signal processing, and not with real time control. The only way they can approach real-time performance is by massively increasing the processing speed, and then you pay for a lot of extra power. And as that power becomes available, the general purpose operating systems always find new ways to fill it up. Your only hope is when you can add on a peripheral that runs it's own real-time os in conjunction with the main system. (which is what Chromatic does with Mpact media processors, and presumably what digidesign does with their system.) So this problem isn't likely to go away soon on basic desktop machines. The hardware based loopers do a lot more than just put a sound in memory and repeat it. What makes them special are all the other functions and features that let the user interact with it like a musical instrument. These are the sort of features that will be very difficult or impossible to create on general purpose machines available now or for the near future. And then there's the longer view of where PC's are going, which may offer more possibilities. A lot of people in the PC and consumer electronics industries are pushing these areas to merge together. So PC's begin to appear that are more specifically focused on single tasks, and consumer electronics devices begin to appear with more PC-like features. You can see this happening now with set-top boxes, PDA's, and home theater equipment. You will soon see the same with stereo equipment, TV's, telephones, dishwashers, and whatever. It's not unreasonable to think the same will happen with musical instruments. I tend to think that in 10-15 years, the idea of using a desktop computer designed for office use for the purposes of creating music will seem as quaint and old-fashioned as the huge patch cord based analog synths of the 60's and 70's seem today. Time will tell I guess. kim ______________________________________________________________________ Kim Flint | Looper's Delight kflint@annihilist.com | http://www.annihilist.com/loop/loop.html http://www.annihilist.com/ | Loopers-Delight-request@annihilist.com