Re: [yoshimi-user] About version 0.062-pre2 - timestamp and seqfaults
- From: detox.maestro@xxxxx
- To: cal <cal@xxxxxxxxxxxx>
- Date: Tue, 16 Nov 2010 23:06:08 +0100
And the last thing is that in this version still don't check timestamp.
Are you sure about that?
Yeah, I think so, maybe it is not the problem of timestamp checking (but
probably).
But at the end I can hear midi jitter - this is hearable clearly at
bigger jack periods sizes.
So this version also have midi jitter as yoshimi version 0.058 or
zynaddsubfx.
This problem is still don't resolved in yoshimi.
For me, the midi jitter issue is a continuing annoyance, but not
actually a
significant "problem", as you seem to find it. I have to wonder what size
jack period you're running.
Yeah, I used max bigest jack period size as possible for testing.
It was just for testing.
Maybe in future versions of LV2 zynadd will have all features than
native zynaddsubfx, who knows?
I think I know the answer to that one.
What is that answer?
Do you have any news from zynadd developers?
Will they continue developing zynadd?
I think that's an incredibly generous invitation for me to spend NNN hours
analysing code and experimenting in order to solve an issue that you
perceive to be a showstopper, but I don't. When the work demands it, I run
jack With a period size of 64, or even 32, whereupon the midi jitter
simply
isn't a significant issue for me. It still troubles the purist element of
my nature, but it really isn't a significant problem.
Yes but even if you use very small period size then it doesn't means
that midi jitter disappear.
Even when I use 256 period size if I hear record carefully then I can
hear midi jitter.
It is hard to imagine for me how to do good record even with small midi
jitter.
You always know that even if you don't hear it the midi jitter is
present in your record.
And remeber also the thing that not everybody have quad-core 3Ghz.
Sometimes it can be something like single core 1Ghz.
In that case it will be hard to play many notes at the same time by
soft-synths with 64 period size.
They will need to use 256 or even 1024 period size.
So I think that this issue is important to resolve.
Now here's an idea! How about _you_ study the zynjacku codebase and
give us
your analysis of how to solve this? Given what I already know about LV2,
plus
the fact that I've looked through the zynjacku code a couple of times over
the last 12 months, I think I can predict where that will lead.
Yeah, it is great idea.
I can study zynadd codebase, but I'm not expert in midi area, sound
developing and in how zasfx or yoshimi is made.
So if you told that it will take for you XXX hours - then for me it will
probably take XXXXXX hours... :(
Other related posts: