[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Date Index][Thread Index][Author Index]

Re: Real-time category

Matthias Grob wrote:
> So if you want to record a full attack, you need to press the button 
> about 7ms before you play it - which does not seem to be a problem for 
> anyone.

When I record into software, I delay the audio signal by the fadetime, 
and can then even hit exactly on the point without missing any attack. I 
am recording always, it never stops, when I want it to go into a loop, I 
route the delayed signal to the place I need it. This artificially 
introduced latency never hurts, but I need less practice for being on time.
As instead of me learning how much I have to be in advance for a 
trigger, my software learns how much I am usually late if I feel I am on 
the "groove"...

> you are exagerating. 

I know, ;-)

> to trigger with audio is not that simple. Andy and 
> me spent quite some time to figure it out for the Mathons plugins. Its 
> never within a sample. And we want foot operation, and I doubt that a PC 
> can garantee 1.5ms for that. But there is no need. Praxis counts!

I think if you just deliver a dc (a switch with a battery), even to an 
audio interface which does not pass dc itself, you should easily detect 
this loud click and use it as a trigger.

You are right, if you use something like a drum sound to trigger, it 
takes a handfull of samples. But even then, if you know roughly the 
waveform in advance, it should only create latency, not jitter.

But I agree, there is not much need, I never felt that my triggers where 
too jittery...

> I dont think so. the 20us go for a RTOS when its clear that changing the 
> task is needed. On the Powerbook there are loads of threads running 
> already and there must be some priority avaliation that finially decides 
> to care for the MIDI event, which then resides in some buffer and so 
> on... I have no idea how long it takes, but I doubt you get to nano 
> seconds.

Its probably still some microseconds, I agree, but its interupt driven, 
that means it will interupt ongoing other tasks of the operating system 
to receive it, get a timestamp, and then switches back. But I am sure 
its much less than the duration of one Midi event, which is roughly 1 ms.



  [][]  [][][]  [][]  [][][]  [][]  [][][]  [][]  [][][]

         Stefan Tiedje
     Electronic Composition

     \\\   /|() ()|\
     ))))) )|  |  |( \\\
     ///    \ \_/)/ )))))
             \___/   ///

--(_|_ ----|\-----|-----()---
-- _|_)----|-----()----------

14, Av. Pr. Franklin Roosevelt, 94320 Thiais, France
Phone at CCMIX   +33-1-49 77 51 72