yes wrote on 2026-09-09, 23:01:

Finally giulio and TUS released 88emu sources at GitHub today.

Cool!

yes wrote on 2026-09-09, 23:01:

I've built binaries for my system (Linux x64) and tested them for a while.

I noticed the cmakelists in the Gearmulator repo references macro's that are only defined in the JUCE repo. Like "juce_add_binary_data".
So is JUCE to be built first, and then somehow cmake can pick up these macro's for building Gearmulator?

yes wrote on 2026-09-09, 23:01:

Finally giulio and TUS released 88emu sources at GitHub today.

Yes, we released the source code, but the software is still in "early access alpha" for now.
You are all of course free to explore the code and build it (it's GPL).
But I kindly ask you to take this version as "unfinished" when it comes to bugs and performance, and wait before doing public "reviews" on blogs/youtube and such (give us time to get it right).
And if you want an already compiled build, or just support us, we accept donations and we have a Discord server where we give updates and progress info: https://theusualsuspects.io/

By the way, if you find problems with 32 channels MIDI feel free to share them! Would love to try them out and fix it 😁

(and of course as soon as the early access alpha is over binaries will be available on our site)

Most 32 channel midis i've seen are composed for sc8 or sc88 pro. running them on the wrong synth may or may not give differences, depending on if they manually request the older maps.

Do report any midis that seem wrong on all three emulations.

yes wrote on 2026-09-09, 23:01:

Finally giulio and TUS released 88emu sources at GitHub today. […]
Show full quote

Finally giulio and TUS released 88emu sources at GitHub today.

5 devices are emulated: SC-88 - pure one, VL and PRO versions, SC-8850 and old good SC-55mk2 - judging from remarks, they used Nuked SC-55 sources for it.
SC-8820 isn't supported atm, but once TUS have the hardware, they will fix it for sure.
Also you have to provide your own ROM images for any device to start.

I've built binaries for my system (Linux x64) and tested them for a while.
88emu appears to be a standalone version, publishes 3 new MIDI IN ports to the system (A, B, announcements) so it should be directly compatible with DOSBox and such, didn't test it yet.
Incorporates a built-in MIDI file player for testing , GUI is scalable from 50% to 300%.

Generally all synths are working but we need more testing to detect alpha version peculiarities.
For example, 32 channel midis sound kind of weird but LCD graphic demos look all OK.

More teasing screenshots to follow:

The attachment 88.png is no longer available

The attachment 88VL.png is no longer available

The attachment 88pro.png is no longer available

The attachment 8850.png is no longer available

The attachment 55mk2.png is no longer available

It's fantastic, WOW!
I hope you also add the SC-55 mk1 (ROM version 1.21) for games that require capital tone fallback.

Oh. have to clone gearmulator... understand now.

no, i don't. how does one actually build this in a current Visual Studio Community Edition?

You use CMake to build it? I got it successfully building after doing a shallow recursive submodule clone with VS2022 build tools using CMake.

MSYS2 Clang does not work yet, evidently.

We'll provide builds as long as we release the next version, so you can also wait

Okay i THINK i figured this out.

I'll know when it's finished.

This is crazy confusing.

NOpe stil dont' get it.

going to just wait.

The recent developments are now getting ported MAME.
At least, that is my understanding. The commit-credits mention GiulioZ.

Sound Canvas SC-88 / 88Pro / 8850 support
https://github.com/mamedev/mame/pull/16080
https://github.com/mamedev/mame/pull/16086

AFAIK MAME SC-55 is in the same Preliminary status for years now:
Driver Status: Preliminary / Sound Status: Unemulated

Very interesting. I presume mame will also be using nuked as a reference to finish their emulation?

still curious what 32 channel midis people are saying sound wrong.

Everything i've thrown at it seems just fine. including the splash cymbal filter test.

btw, this player doesn't provide access to the power button. this means to send a gs reset after poweron to fix the mkii pitch issue, you must use the options menu to do the reset instead.

The work in mame is from superctrl, not from me, it uses a completely different h8 and xp emulation core

I think this also belongs to this thread. At least I'm very excited 😀
DOS enabler + host for Yamaha PCC10XG PCMCIA MIDI card https://github.com/zikolas/pcc10xgo
Makes it usable on standard 330h MIDI port. Testing and feedback is appreciated

S-MU2000 emulator dropped.

somewhat challenging to build, but it's contributing back to mame.

and it's GOOD.

This project's aim is to make a VST3 instrument for DAWs, btu it also generates a guiless midi listener, as well as a gui player that can also listen on virtual midi ports.

It akes MAME roms, EXCEPT for the control rom, which it has you extract from yamaha's firmware updater using python.

It also provides a dumper custom firmware you can flash to real hardware to then dump the wave roms over usb or even midi out.

You then of course re-install the 2.01 formware update to fix your real device.

I'm waiting on mod OK to post my simple exe tool that coverts the mame control rom to the format this emulator expects, and places files in the correct subdirectories.

Meanwhile the guy that added the Mame Sound Canvas SC-88 / 88Pro / 8850 support, also promoted the SB-55 to working Roland Sound Brush that is.
https://github.com/mamedev/mame/commit/a5da980

FluidSynth 2.6.0, released this august, also introduced some noteworthy features:
- Add new reverb engines to fix flaws of fluidsynth's previous FDN reverb
- Fluidsynth's default reverb engine was changed to Dattorro - a plate-style reverbator
- Fluidsynth now uses a higher order sinc interpolation
I can still build it for Windows XP no-problem 😀

Earlier shingo45endo published the browser/web-based sc55mk2-ctf-patcher.
I noticed he has a new project at GitHub:
https://github.com/shingo45endo/sc-mcu-wave-romdumper
Which should make it easier to extract ROM's for synth emulators from their original hardware. Again browser/web-based.
"Dump the wave ROMs (and the MCU's internal ROM) of a Roland SC series over MIDI, all at once."

giulio wrote on 2026-09-10, 14:42:

(and of course as soon as the early access alpha is over binaries will be available on our site)

Does anyone know when the early access alpha is done and binaries are put on your website ?

The emusc fork by Marc Lainez, now called emuscd, is going in its own direction now, separate from the main emusc project.
https://github.com/mlainez/emusc

About This Fork This fork exists because exploration of the SC-88 emulation required detailed ROM analysis, measurement of envel […]
Show full quote

About This Fork
This fork exists because exploration of the SC-88 emulation required detailed ROM analysis, measurement of envelope behavior, and reverse engineering of oscillator characteristics. This work was substantially assisted by Claude (an AI assistant), which helped:

Analyze ROM structures and extract device profiles
Test and validate TVA envelope curve fitting
Implement parameterized oscillator stages
Trace MIDI voice allocation behavior
AI Involvement: Following the kernel.org coding-assistants convention, all commits carrying AI assistance include an Assisted-by: Claude:<model> trailer. The human author retains full responsibility and is listed as the author of record.

Why separate from upstream? The upstream EmuSC project focuses on a polished GUI application for end-users. This fork prioritizes library development, low-level analysis, and exploration driven by reverse engineering. The two projects have different goals and audiences.

His SCVA-Headless project is also being updated.
Like "Let scva-winmidi change the map from the console while it runs"
And something about Windows 98 Support!?
Noticed this new documentation snippet:

The core imports 50 functions: 26 from KERNEL32, 9 from VCRUNTIME140 and 15 from the Universal CRT. Windows resolves those through its own loader, so the Visual C++ redistributable is a real requirement there and nothing else is. Linux never sees it, because pe_loader.c supplies the imports itself.