VOGONS


First post, by davidhs

User metadata
Rank Newbie
Rank
Newbie

Hello all, I'm David. I seem to have wandered around the forums but I think now is the best time to create a new account in here 😁

I tried to modify Munt to use the Roland D-110 ROMs dumped on the net and surprisingly it works quite well!

Although I find a few differences from the real one that I have, for example the AcouPiano1 tone in the real module seem to have shorter decay than the one in Munt.

I posted a YouTube video on demonstrating it, playing it's own demo song: Macho Memory:

https://www.youtube.com/watch?v=pyk_pJlwqiU

Reply 1 of 12, by Spikey

User metadata
Rank Member
Rank
Member

David, this is freaking cool! I had asked about something like this, but never got a response.

As a fellow D-110 user, I would be VERY appreciative if you could send this to me/upload it!

Reply 6 of 12, by sergm

User metadata
Rank Oldbie
Rank
Oldbie

From a quick look, the tune starts with a 16-byte title, then goes 16 bytes of system area (?) and the 256-byte patch dump, and then a stream of timestamped MIDI messages likewise from a SMF track go afterwards.

Reply 7 of 12, by davidhs

User metadata
Rank Newbie
Rank
Newbie

Thanks for all the positive response.

This is actually something I made because from what I knew, D-110 have very similar synth structure as an MT-32, so I gave it a try.

Turns out it works quite well, and it sounds beautiful as well. Although to make it the most complete, I still don't know how to parse the reverb chip ROM dump I found in the Internet (I'm sorry I forget what is the name of the file), so that the reverb settings are at most accurate from the device itself.

I also need to modify the current Munt code to support D-110 "patches", i.e. the reverb and partial reserve settings.

Reply 9 of 12, by davidhs

User metadata
Rank Newbie
Rank
Newbie
sergm wrote on 2020-10-29, 19:24:

From a quick look, the tune starts with a 16-byte title, then goes 16 bytes of system area (?) and the 256-byte patch dump, and then a stream of timestamped MIDI messages likewise from a SMF track go afterwards.

Hi, sergm!

The 256 byte data in the start seems to be some kind of patch map settings, but I'm still not sure what setting it sets.

Reply 10 of 12, by davidhs

User metadata
Rank Newbie
Rank
Newbie
krcroft wrote on 2020-09-29, 15:35:

@davidhs - I like what I hear. This is a very exciting addition to the Munt project!

Do you by chance have have source repo where you can share your updates?

Thanks, krcroft!

I have a github account, but the changes aren't uploaded to my repo yet, as I'm still currently busy WFH..

I also still need to implement the D-110 "patches" memory and I also need some help parsing the reverb settings from the ROM dump, because it has more reverb settings than the MT-32.

Reply 11 of 12, by sergm

User metadata
Rank Oldbie
Rank
Oldbie
davidhs wrote on 2020-11-02, 17:50:
Thanks for all the positive response. […]
Show full quote

Thanks for all the positive response.

This is actually something I made because from what I knew, D-110 have very similar synth structure as an MT-32, so I gave it a try.

Turns out it works quite well, and it sounds beautiful as well. Although to make it the most complete, I still don't know how to parse the reverb chip ROM dump I found in the Internet (I'm sorry I forget what is the name of the file), so that the reverb settings are at most accurate from the device itself.

I also need to modify the current Munt code to support D-110 "patches", i.e. the reverb and partial reserve settings.

Well, adding support for D-(1)10 synth requires quite a bit of changes. There are more ROMs involved, more timbre banks, etc. And I'm completely unsure whether the MT-32/CM-32L emulation model is remotely correct. Studying that requires an effort. Perhaps, easier is to go another way and emulate the MCU fairly for other synth. Really, today computers are even faster, so full-blown emulation may be more desired being accurate but slower. For instance, MAME/MESS already has implemented the MCU model (albeit the sounds hardware is still missing 😒). I guess the speed is more valuable for gaming, so probably mt32emu will remain a HLE (e.g. the measured speed of reverb emulation was 20 times faster in mt32emu when I tested).

Also, speaking of reverb, if you are interested in how the reverb chip presumably works, don't hesitate to look at the BossEmu thingy. It is already capable to work with some more reverberation stuff other than MT-32 and the likes. This is probably most important difference comparing to the mt32emu emulation model, as the D-(1)10 units have the number of reverb modes doubled, AFAIK 😀