DracoNihil wrote on 2024-07-28, 04:38:

What I don't understand about the infamous "GM.DLS" is that, after hearing a real SC-55 and this emulator and then looking at th […]
Show full quote

zaphod77 wrote on 2024-07-28, 03:54:

That is a low quality sc55mkII (mostly) sampleset. and not suitable for use in an actual SC emulation.

What I don't understand about the infamous "GM.DLS" is that, after hearing a real SC-55 and this emulator and then looking at the ROMs you have to use... how is it that "GM.DLS" is the filesize that it is yet it seemingly doesn't have all of the PCM samples?

The combined size of both "waveroms" is 3 MB and "GM.DLS" is 3.3 MB so I don't understand how most of the instruments in that DLS don't have all of their PCM samples present, or is there something else going on in the SC-55's PCM chip that is producing the sounds I don't hear "Microsoft Synthesizer" doing.

For context: I'm talking about the "String Ensemble" program. Especially when played at the lower (below "Middle C") octaves.

Depends how the samples are compressed. Roland uses 2:1 companding so the true size is 6 or so MB. Although minus the MT-32 sounds (1/2 MB), it's 2/4.
So 3.3MB, depending what level of compression is present, may not be the full 4-6 MB.

But the key reason it sounds garbage is the lack of effects. Roland's warm reverb and thick chorus absolutely transform the SC-55 sound.

yeah, the sample rate of gm.dls is low, and that's why it's so small. Roland even said to me when i asked if it was okay to convert it to an sf2 for personal use that they were low quality samples. as i've mentioned there answer was, yeah that's fine but don't distribute it.

GM.DLS also sounds to me like they're using fewer PCM samples to cover more octave ranges instead of using more samples recorded/generated at different pitches which would obviously lead to some poor sounding notes.

MusicallyInspired wrote on 2024-08-27, 15:19:

GM.DLS also sounds to me like they're using fewer PCM samples to cover more octave ranges instead of using more samples recorded/generated at different pitches which would obviously lead to some poor sounding notes.

This is EXACTLY what I was talking about earlier. I don't understand why GM.DLS was engineered like that.

Makes sense to me given it's an OS component that must load into memory on demand and Win98 targeted 16-32MB systems. MS had all the opportunities to improve on it since then but didn't. 🙁

But also as with low-res pixel art in the late 90s, there was a similar sentiment about 'beeps and bloops it sounds like a midi!!!!!! its not 80s anymore' regarding mainstream OPL synthesis so anything's better than that I guess!!!

leileilol wrote on 2024-08-27, 19:50:

But also as with low-res pixel art in the late 90s, there was a similar sentiment about 'beeps and bloops it sounds like a midi!!!!!! its not 80s anymore' regarding mainstream OPL synthesis

In the PC scene at least, that sentiment was initially and chiefly driven by Creative's AWE marketing, as I recall.

zaphod77 wrote on 2024-07-28, 18:12:

yeah, the sample rate of gm.dls is low, and that's why it's so small. Roland even said to me when i asked if it was okay to convert it to an sf2 for personal use that they were low quality samples. as i've mentioned there answer was, yeah that's fine but don't distribute it.

GM.DLS is also licensed out by Roland for use in console games, so keeping it small and simple was part of the design.

Shreddoc wrote on 2024-08-27, 22:01:

leileilol wrote on 2024-08-27, 19:50:

But also as with low-res pixel art in the late 90s, there was a similar sentiment about 'beeps and bloops it sounds like a midi!!!!!! its not 80s anymore' regarding mainstream OPL synthesis

In the PC scene at least, that sentiment was initially and chiefly driven by Creative's AWE marketing, as I recall.

Looking at the quality of the tiny AWE32 ROM font, "anything better than OPL is fine" is the correct sentiment to sucessfully market that card. To be fair, for an uncompressed 1MB wavetable ROM, the sample set is not that bad, though.

Issue 92, brings up somthing I've been thinking about.

Now, with this perfect emulation, and all thees firmware files being dumped.

Can't a new firmware be created that fixes thees bugs try community bowling knows about. Like the removed fall back feature, let's add that back in, in later versions.

Crate the ultimate version that has all the features and now bugs.

With emulation, testing becomes easy, and ensures the fix would work on real hardware.

Karmeck wrote on 2024-09-04, 21:45:

Issue 92, brings up somthing I've been thinking about. […]
Show full quote

Issue 92, brings up somthing I've been thinking about.

Now, with this perfect emulation, and all thees firmware files being dumped.

Can't a new firmware be created that fixes thees bugs try community bowling knows about. Like the removed fall back feature, let's add that back in, in later versions.

Crate the ultimate version that has all the features and now bugs.

With emulation, testing becomes easy, and ensures the fix would work on real hardware.

Games that use CTF probably work fine on the original SC-55. For those that were composed on SC-55 mk2 and later, well... I guess the composers used the instruments correctly, using the correct Program Changes since they couldn't make mistakes that would then be "fixed" by CTF since it was no longer there. This is just my guess.

There is an example with System Shock which was probably never tested on Roland synthesizers without CTF. This game sets the drums with a wrong Program Change that SC-55 mk2 and later models do not recognize and return an error. Here the problem, more than the lack of CTF, is the incredible mistake of the composers and I could also hypothesize that CTF hid the mistake (or did they get the Program Change wrong on purpose because other synthesizers required it and so on Roland synths they relied on CTF?).

I guess if someone composes a track correctly on SC-55 mk2, he does not need CTF (right?).

mkarcher wrote on 2024-08-29, 15:48:

Shreddoc wrote on 2024-08-27, 22:01:

leileilol wrote on 2024-08-27, 19:50:

But also as with low-res pixel art in the late 90s, there was a similar sentiment about 'beeps and bloops it sounds like a midi!!!!!! its not 80s anymore' regarding mainstream OPL synthesis

In the PC scene at least, that sentiment was initially and chiefly driven by Creative's AWE marketing, as I recall.

Looking at the quality of the tiny AWE32 ROM font, "anything better than OPL is fine" is the correct sentiment to sucessfully market that card. To be fair, for an uncompressed 1MB wavetable ROM, the sample set is not that bad, though.

IMHO, that 1MB AWE32/AWE64 ROM +EMU8000 FX engine generally sounds better than the Windows GS softsynth abomination.

darry wrote on 2024-09-05, 06:46:

IMHO, that 1MB AWE32/AWE64 ROM +EMU8000 FX engine generally sounds better than the Windows GS softsynth abomination.

That abomination was sold under the name Edirol Virtual Sound Canvas so it wasn't that bad 😆. The problem is that the free Windows version doesn't support effects like Chorus and Reverb.

Apart from the low-quality DLS file used to power it, the Microsoft Wavetable Synth was never really feature-complete software or suitable for use. It just barely worked when it was introduced and could handle basic GM files, but wasn't GS-compatible in the slightest, despite being inferred as such, in the press releases and naming. The reason why there was no reverb/chorus is because all effects are done through DirectSound and you'd have to have a DirectMusic/DirectSound-compatible MIDI player to get that, by calling up the effects in DirectSound, when triggered by the DirectMusic API. The only one - that I know of - that really ever existed was what was in DirectProducer, for making tracks with DirectMusic, so you'd never get it through any regular MIDI player (Windows Media Player and others) that used the Windows MIDI stack to run MIDI through the Microsoft Wavetable Synth.

Never late to say THANK YOU, @nukeykt

What's new coming up? A built-in driver for Windows, a graphical interface to adjust settings, or an icon for the program?

Kappa971 wrote on 2024-09-10, 11:25:

What's new coming up? A built-in driver for Windows, a graphical interface to adjust settings, or an icon for the program?

Going off github fork and pull request activity. Internal restructuring and wave-file output by jcmoyer. ASIO support and support for more Sound Canvas devices and related synths by giulioz.

markanini wrote on 2024-09-10, 13:09:

Kappa971 wrote on 2024-09-10, 11:25:

What's new coming up? A built-in driver for Windows, a graphical interface to adjust settings, or an icon for the program?

Going off github fork and pull request activity. Internal restructuring and wave-file output by jcmoyer. ASIO support and support for more Sound Canvas devices and related synths by giulioz.

Looking at Twitter, nukeykt posted videos of a, work in progress, gui. This was back in May.