Hi Guys,
I posted this over at TAKP today, I figured I would post it here as well. It seems like Lilcea was down the same path as I was.
Just stumbled across this thread and thought I would share some thoughts and findings on this topic, as I have been researching it for a while.
A little history first - I played EQ in 2000 on a Win98 box with an AWE32 and remember how great the music sounded with this hardware. For years I've looked for ways to get the music sounding as I remember.
SO in the case of the EQ client itself - As mentioned previously; about the time when PoP came out, they removed the clients ability to dynamically switch the soundfont banks into EMU8000 cards. This had to do with them trying to fix the compatibility issue with Windows XP/NT and Creative drivers no longer allowing access to bank switching directly by applications. Windows 9x/DOS allowed for much lower level access to the hardware through the drivers. If you tried to use any period EQ client on a Windows XP/NT/2000 machine, soundfont bank switching would NOT work. Using a Trilogy or Luclin era client on a Win9x box with an AWE32 card and correct drivers should dynamically switch in the proper banks automatically.
Now let's dive into the meat and potatoes of what I've found. Using VirtualMIDISynth, Timidity++ etc. and loading the banks 1mgm.sf2 and synthusr.sf2 in the appropriate manner gets us CLOSE to what it sounded like, but it's not exact. This is because of how synthusr.sf2 relies on the onboard AWE32 ROM (in this case 1mgm.sf2) to patch in missing wavetables. This can be verified by opening synthusr.sf2 into a soundfont editor like Viena.
[You must be logged in to view images. Log in or Register.]
Here you can see synthusr.sf2 loaded into Viena. Notice on the left side under "samples" we see a list of !ROM: samples. These are the samples that synthusr.sf2 is expecting to import from the AWE32 ROM. If we click on one, we get an error that it isn't found.
Now let's load up 1mgm.sf2 into Viena - and what do we find? All of the samples synthusr.sf2 is looking for, of course.
[You must be logged in to view images. Log in or Register.]
Back to synthusr.sf2 - Now let's look at the instrument presets, and we'll notice that not only do many of the presets use a combination of wavetables - some from synthusr.sf2 and some from 1mgm.sf2, but synthusr.sf2 applies specific parameters to the wavetables from 1mgm.sf2 that cannot be reproduced by loading the soundfonts into a software emulation.
For this example, let's take Brass 1 from synthusr.sf2
[You must be logged in to view images. Log in or Register.]
We can see that this instrument preset combines BrassSectC3 from 1mgm.sf2 4 different times to create a unique sound that's different from just using the preset from 1mgm.sf2 by itself. Every single preset in synthusr.sf2 uses the AWE32 ROM samples in this function. This is also true for all of the drum presets. The limitations of software like VirtualMIDISynth doesn't allow individual wavetable samples to be imported in these cases.
Now what we can do is import and reassign every instrument preset in synthusr.sf2 to use the correct samples, then save the resulting .sf2 out and in this case load it into XMPlay with an original EQ .xmi file. (note: This new .sf2 removes the need to load 1mgm.sf2 at all)
The results speak for themselves - Here are a few tracks using the new .sf2 with correct samples:
Akanon track 1:
https://sndup.net/2kts
Qeynos tracks 2 and 6:
https://sndup.net/5wfj
https://sndup.net/9gn5
West Karana (qey2hh1.xmi track 3)
https://sndup.net/853w
Kaladim tracks 5 and 6 (notorious for sounding terrible with the wrong soundfont)
https://sndup.net/536w
https://sndup.net/59m3
I think you'll notice the extra layers to instruments you'd normally be missing along with the correct percussion sounds. The pieces now sound as close as to what I can remember and how Mr. Jay Barbeau originally intended.
If you want to give it a try, the new .sf2 file is
here (I highly recommend XMPlay, as it has built-in soundfont support and you can load up the original EQ .xmi's with all of the subsongs).
As a last note - There is one other variable not taken into account - The EMU8000 on the AWE32 also allowed for hardware effects processing through EAX, which could be applied to soundfonts. I am not 100% sure about how the EQ client went about this, but if someone does get a Win9X box running with a Trilogy or Luclin client and AWE32, it would be interesting to compare results.
Edit: I have dumped all of the .xmi subtracks to .mp3 using this new soundfont. You can find the .zip of them all
here. (633 MB)