Right, I should be able to help out with this shortly. I have a number of devices I can offer, but I imagine most of them have already been covered.

- SC-55mkII
- SC-88ST
- SC-88ST Pro
- SC-8850
- MT-120
- PMA-5

An SC-33 is also arriving next week.

One thing I must say is that although the demo tracks are indeed very well programmed, they don't fully exploit the differences that can be heard between different engines employed in the devices. For example, a rendition of Michael Jackson's Thriller (which sadly doesn't have any author information embedded) employs some clever filtering on both the brass sections (to make them brighter and snappier) and on the synth sections (to make them sound more analog) and this sounds vastly different across modules.

That being said, having re-read the first post, this project mainly focuses on the DAC rather than the actual device engine quirks and features.

Just a little feedback anyhow - I'll try and get these recordings done at some point soon for you.

And for anyone who is curious about the track I mentioned - please see attached. I've edited the track to remove the vocal tracks as they always sound terrible, no matter what song.

As a fact, i can confirm, that the MT-32 analog recordings looks almost (99,99%) the same as digital recordings from my MT-32 SPDIF mod.
They are not bit-to-bit identical, but this is also about different analog line in level, which i cannot set to equal digital in.

firage wrote on 2022-03-12, 12:43:

Sounds very interesting. I'd be happy to participate with my SCC-1 sound card in any way that's helpful. 😀

Recording gear and methods will impact the audio samples, so it'll be good to attach some of that information to all the recordings.

I did not read through this whole thread, so I apologize if this was already mentioned, but there are two versions of the SCC-1; the SCC-1A and SCC-1B, which the specs look a lot like the earlier SC-55 (not fully GM compatible) and SC-55mkII.

There are also different firmware versions for otherwise identical synthesizers, which could produce a different output. So for any submissions, the firmware version should be included as well.

See here for some research I did on all of this previously:

Significant differences with certain MIDI files across Roland synthesizers

Kahenraz wrote on 2024-08-21, 18:27:

I did not read through this whole thread, so I apologize if this was already mentioned, but there are two versions of the SCC-1; […]
Show full quote

firage wrote on 2022-03-12, 12:43:

Sounds very interesting. I'd be happy to participate with my SCC-1 sound card in any way that's helpful. 😀

Recording gear and methods will impact the audio samples, so it'll be good to attach some of that information to all the recordings.

I did not read through this whole thread, so I apologize if this was already mentioned, but there are two versions of the SCC-1; the SCC-1A and SCC-1B, which the specs look a lot like the earlier SC-55 (not fully GM compatible) and SC-55mkII.

There are also different firmware versions for otherwise identical synthesizers, which could produce a different output. So for any submissions, the firmware version should be included as well.

See here for some research I did on all of this previously:

Significant differences with certain MIDI files across Roland synthesizers

Sure, there are ROM versions for all of this stuff. Minor firmware changes don't usually make a difference unless the material is specifically picked to bring it out, though it's always interesting when you do find something.

I know a major revision exists that's marked "SCC-1A "on the PCB. (I believe "SCC-1B" is just an SKU for the SCC-1A.) Mine is an original SCC-1 without the A. From my testing: "Equivalent to SC-55 version 1.2x plus an intelligent mode MPU. Features Capital Tone Fallback and drum channel program changes also fall back to nearest like early SC-55's. GM standard configuration of PC#122 Breath Noise / Fl. Key Click, but doesn't have the additional instruments from SC-55mkII." I can tell you that the top of the Program ROM chip reads "GSS.PRM.A 2, R15279809".

I read that it's not fully GM compatible, because it interprets the GM reset command as GS reset, like the other early SC-55s, can you confirm? I think that this is a problem for all firmware prior to 2.xx.

Kahenraz wrote on 2024-08-22, 12:14:

I read that it's not fully GM compatible, because it interprets the GM reset command as GS reset, like the other early SC-55s, can you confirm? I think that this is a problem for all firmware prior to 2.xx.

I'll have to re-check whether it flips Rx.NRPN on GM reset, but I'm quite sure it's just a GS reset. Because that's how the SC-55 originally worked, games do use GS features despite sending GM resets. Strict GM devices are more trouble then.

What does that mean? More trouble for newer devices?

I mean depending on your synth's reset behavior one game will sound as designed and another will not. The original SC-55 behavior was de facto standard for a couple of years.

On a seemingly random notion, I just installed the last version of Quicktime For Windows on my computer and the MIDI (Roland GS soundset) works wonderfully. It kicks the crap out of the Microsoft Synth and it's DLS file. It doesn't match up to the SoundCanvas VA, but it's pretty passable with GM and basic GS files, so maybe it should be added to the comparison list.

Nice comment. Didn’t know nor remember that QuickTime would install midi drivers. Thx.

I might add 2 more contestants: Roland EM-20 and Roland AT-90. As far as I can tell, AT-90 uses SC-33 samples for GM mode, while EM-20 has SC-88 drums and cymbals.
https://youtube.com/shorts/U3dB9QxQRLk?si=Q5NRLY8gLh7tHMxf
https://youtu.be/SqzRvrQFxpQ?si=yA1OlR384n0qgTko

Yeah, basically hundreds of Roland keyboards used either direct Sound Canvas sample sets, or supersets of them - some new ones still have the SC-8820/50 set but expanded. There is probably 200+ keyboards that would meet the criteria, but nearly no-one in this space will have them.

I think all the Roland EM boards were Sound Canvases. The EM-20 is some version of a SC-55 I think, possibly.. After checking, it's in fact a hybrid of the SC-55, 88, 88Pro and even MC-303 totalling an 8MB sample set (versus 2MB for the usual SC-50 or '55 minus MT-32 tones').. weird. I think something similar would be used in some of the MT "Sound Canvas speaker" units, for reference.. Making a note of that. I think one of them has the same sound set.. MT-90U or similar.

The AT-90 appears to be a SC-50 GS part married to a different sound set of unknown origin, at least to me. Interesting.

Roland still uses a basic GS offering on it's home/studio pianos, for the purposes of MIDI file accompaniment and composition, but I'm pretty sure that it's been gone from their production and performance keyboard line for many years now, probably since the early 2000's. Yamaha also still has their basic XG implementation on some of their equivalent products, which also includes the emulation of basic GS that XG offered through the TG-300 and others.

RetroGamer4Ever wrote on 2025-01-27, 19:22:

Roland still uses a basic GS offering on it's home/studio pianos, for the purposes of MIDI file accompaniment and composition, but I'm pretty sure that it's been gone from their production and performance keyboard line for many years now, probably since the early 2000's. Yamaha also still has their basic XG implementation on some of their equivalent products, which also includes the emulation of basic GS that XG offered through the TG-300 and others.

Looks like Roland's E-A7 current Arranger keyboard still uses almost all of the SC-8820/50 instruments in its patch map, for reference - some GM patches derived from the SC-55 I would imagine.

GS specifically isn't offered, seemingly replaced by GM2, which is mostly GS instruments being included in the official standard anyway. Although I think all the instruments are there, just not mapped to GS in that sense.

Looks like the Roland BK-5 arranger keyboard is also heavily using SC-8820/50 tones, and IS explicitly GS compatible still.

Actually, I just discovered that both of the products you mentioned are full GS synths, though they are not advertised as such in the specs or use the GS branding anywhere on the body. The MIDI implementation document for both says that the default operating mode for both products is GS, with GM1 and GM2 as optional operating modes.

Well, we both learned something today!

As far as bringing it full circle, I'm not interested in including every random GS keyboard, and it's not really what old gamers care about in any case. Some exceptions for keys like the JV-50 (literally a SC-50 keys with JV expandability, but possibly with some SC-88 tech as well), SK-50 (SC-50 keys but with 55 output stage/DAC), and so on.

If people submit recordings of some, it's fine to include it, but the goal is not to include every one by any means. Most pre-2000's ones are equivalents of hardware modules, the variations being of more interest like the examples mentioned above, and post-2000 are random arranger keyboards with modern for the era tech that no longer relate to the SC core set, and that most people wouldn't be interested in, or would have trouble, acquiring.

While newer products like the SD-50, etc also fit that bill, they're modules/desktop units and not huge keyboards.