Re: [yoshimi-user] Different sound in Yoshimi and ZynAddSubFX.
- From: "Mark D. McCurry" <mark.d.mccurry@xxxxxxxxx>
- To: yoshimi-user@xxxxxxxxxxxxxxxxxxxxx
- Date: Wed, 16 Oct 2013 14:46:49 -0400
Just to eliminate any inaccuracies:
1) it crashes very often (maybe because of bad compatibility with
JACK?), especially while turning the knobs and playing at the same time.
Major work has been done in this regard, though I won't pretend that yoshimi
ever properly solved these issues.
If anything yoshimi has turned these jack issues into a fun game of undefined
behavior
2) the GUI is difficult to use (the knobs don't interact smoothly with
the mouse pointer)
Odd, I thought the knob handling code was the same at this point.
If you can define how things are non smooth could you perhaps put an entry up on
the zyn bug/feature tracker?
When one forks a synthesizer, one will produce differences in result, because
the wave generation will change with any changes in the code.
I don't know where you're coming from on this.
Forking any project for synthesis or general processing does not need to change
the resulting output at all.
Sure, there may end up being some decisions to purposefully change things, but
otherwise differing output can generally be defined as a bug.
I would have to use enormously more hardware to achieve the results being seen
right now, if I had to use ZASFX.
If you're talking about the performance differences, this was debunked ages ago.
Different default compiler flags result in different performance, *surprise*.
It doesn't help that distro binaries tend to reduce what can be used for
optimization (eg the elimination of useful sse instructions in favor of a more
generic image).
Also, does anyone know if the sounds produced by Zyn. have changed over time.
Hmmm. Interesting. It seems I have an archived copy of Zyn. 2.2.1. I'll try
and
get that up and running again too.
The sound generation *should* be fairly stable, as I've had tests setup to
ensure
almost sample accurate generation for some time now (almost sample accurate as
floating point math never is quite the same when you introduce -ffast-math).
That said, there were some sizeable changes with unison around that time, so
that could have changed things slightly and also the codebase is huge, so the
tests that are in place only cover so much.
--Mark
Attachment:
pgpjXNbW4N7Nw.pgp
Description: PGP signature
Other related posts:
