>I came across this article (hopefully this isn't old news for most of you) >about the new Steinberg VST System Link and found myself quite excited about >the possibilities (assuming it will work well).
i had mailed something to LAD about this last week, but it never showed up. >I know this is Windows/Mac stuff, but I can't help but think how incredible >it would be to have a linux box be one (or more) of the nodes. It appears >to just use the digital outputs of an ASIO capable device so there is no new >hardware to buy either (assuming you have such a device and the OS to run >it). > >I suspect that it is some proprietary streaming protocol they've come up >with, but I'm not sure. I'm also not sure if Steinberg would be willing to >release that spec. I'd suspect not. i doubt it too. but i've been thinking about this since i read about VST Link, and here's how i think they do it. they say that it requires a single bit of a single audio channel per node. all that needs to happen is to provide sample "clock" sync between nodes. so, you designate a master node. every time the master node processes N frames, it sets a bit in the "interconnect audio channel" for each node that should advance N frames. something like that. the slave nodes watch that channel (notice the possible limit on the number of nodes - this hasn't been documented), and when their bit ticks, they move forward in time. however, this isn't enough to provide temporal sync to a transport control, so its not clear to me how the "single bit" method can work for this as well. after all, the method i outline above is really just a crude word-clock system (low resolution). i'll ask on vst-plugins if they intend to make the technology open. >The point of this post is that I could see JACK doing a similiar thing with >LADSPA plugins on multiple linux nodes. Why JACK? JACK is not a LADSPA host. --p
