[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [fluid-dev] Fluidsynth scale tuning
From: |
Josh Green |
Subject: |
Re: [fluid-dev] Fluidsynth scale tuning |
Date: |
Sun, 21 Jan 2007 13:40:54 +0000 |
Well, as I've mentioned before my knowledge of FluidSynth isn't all that
complete. From an initial look at things though:
Looks like the cents to hertz calculation is where the action is
(fluid_ct2hz_real). This can be found in src/fluid_conv.c. I really
don't like the way that function looks ;) There could indeed be some
issues with the way this conversion is done. There is a table of 1200
values which corresponds to the absolute linear amount of each cent
value from 0-1199 (i.e., 2 ^ (i / 1200)). I'll have a look over it some
more to verify that all is OK. If you feel inspired to look at it
yourself and see something amiss, you know what to do :) Cheers.
Josh
On Thu, 2007-01-18 at 19:54 +0000, Edenyard wrote:
> Thanks for that, Josh!
>
> > I've also noticed rather unpleasant sounding tones sometimes. One thing
> > I have found is that this is often caused by the chorus and/or reverb.
> > Try turning both of those off, or changing the parameters thereof.
>
> I'll give that a try later. It's a great comfort to know that someone
> else has noticed it!
>
> Would you be able to comment on the scale tuning, though? Is it supposed
> to be equal-temperament or something else? Has its accuracy across the
> scale ever been checked? Are the semitone intervals set by means of a
> static table or are they calculated somehow using the twelfth root of
> two or something?
>
> > Current FluidSynth annoyances for me:
> > - Gain control is rather hard to get right (output is often clipping)
>
> I've found that it seems to have a very peculiar 'law': setting it to
> something less than 1 is fine but it suddenly starts to clip when
> increasing above 1.
>
> > - Output clips even when sound card output isn't maxed (internal
> > clipping?)
>
> I've suspected that for a while now.
>
> Thanks for your thoughts.
>
> Cheers, Gerald.
>