Re: [fluid-dev] Son of ticket #65

[Elimar Green]

Hey guys,

Not sure I understand what is going on with this one, but it sounds
like it could perhaps be something I had a hand in breaking ;)
Therefore a feel slightly more responsibility in helping to resolve
it.  Do we know what specific logic changed in FluidSynth and why?
I'm sure there is an SVN log entry that might help explain things as
well as an email archive trail.  It sounds as if its related to the
whole gm/gs mode mess that was never really correctly resolved.

Let me know if I should look into it.  Still traveling about with
intermittent access, but I could make the extra effort if need be.

I just logged into the FluidSynth Trac bug system and noticed that
almost all the bugs are gone.  I can't imagine they have all been
fixed and suspect that instead they were auto removed because they
weren't assigned to anyone.  Hmmm.  I wonder if they can be revived.
The scripts I used to add the bugs are unfortunately not accessible to
me at the moment.

Best regards,

Elimar


On Wed, Jul 28, 2010 at 1:45 PM, David Henningsson
<address@hidden> wrote:
> 2010-07-28 22:21, Pedro Lopez-Cabanillas skrev:
>> On Wednesday, July 28, 2010, David Henningsson wrote:
>>> 2010-07-27 18:34, Pedro Lopez-Cabanillas skrev:
>>>> Hi,
>>>>
>>>> Something similar but not exactly the same to ticket #65.
>>>>
>>>> Sending a bank/program change CC MSB=1, PC=48 corresponding to a GM2/GS
>>>> Strings instrument, FluidSynth changes to percussion bank 128, Orchestral
>>>> set.
>>>>
>>>> Test case: load the GeneralUser GS soundfont [1] and play the demo song
>>>> "March #1 in D major".
>>>
>>> Banknumber is MSB*128 + LSB, in this case 128, there seems to be program
>>> changes for channel 11 to 15 to bank 128 prog 48 in the midi file, and
>>> there is a bank 128 prog 48 preset in GeneralUser_GS, and its name is
>>> "Orchestral".
>>>
>>> What is FluidSynth doing wrong?
>>
>> The formula ("Banknumber = MSB*128 + LSB") is not correct for GS instruments.
>> The bank number is the MSB value alone in this case. So, a MIDI message CC#0
>> val=1 means bank#1.
>>
>> Changing this behavior breaks a lot of GM2/GS songs that now sound bad when
>> played by FluidSynth.
>
> Okay, thats bad.
>
> If we assume it was changed on purpose, changing it back will just
> probably break someone else's songs, I'm afraid. And that the GS/GM2
> specs are closed does not improve the situation, so it's difficult to
> verify. :-(
>
> Any suggestions? Adding yet another setting?
>
> // David
>
> _______________________________________________
> fluid-dev mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/fluid-dev
>