I can understand the usefulness of this for a General MIDI device, but the CM-32L is based on the MT-32, which has a known polyphony. What is an examples are there where the composer wrote music for the MT-32 without taking this into consideration?

does it actually renumber the channels on split, so you said odds, evens so like 1,3,5,7,9 becomes 1,2,3,4,5 on device 1? the screenshot of the sc55 displays seems like this is not the case, so your not actually doubling the channel count, so seems kind of pointless. you could just run a midi through box and get the same result without all the messing around and get all the voices from both boxes instead of half the voices from each box... (and I could be totally misinterpreting yours setup making this reply pointless 🤣!)

My emu soundengine has a dedicated poly out through mode which lets you double the channel count transparently, its very nice. but then you can get a roland soniccell and have 128 voices of polyphony and make everything else look tiny 😀

Kahenraz wrote on 2022-04-27, 12:30:

I can understand the usefulness of this for a General MIDI device, but the CM-32L is based on the MT-32, which has a known polyphony. What is an examples are there where the composer wrote music for the MT-32 without taking this into consideration?

I'm not aware, and I'm just doing it for the fun of it (and peace of mind) 😀

BloodyCactus wrote on 2022-04-27, 13:03:

does it actually renumber the channels on split, so you said odds, evens so like 1,3,5,7,9 becomes 1,2,3,4,5 on device 1? the screenshot of the sc55 displays seems like this is not the case, so your not actually doubling the channel count, so seems kind of pointless. you could just run a midi through box and get the same result without all the messing around and get all the voices from both boxes instead of half the voices from each box... (and I could be totally misinterpreting yours setup making this reply pointless 🤣!)

My emu soundengine has a dedicated poly out through mode which lets you double the channel count transparently, its very nice. but then you can get a roland soniccell and have 128 voices of polyphony and make everything else look tiny 😀

Well, one MIDI channel can be assigned more than one track. Consequently, more than one "voice" can result from a single channel. 16 channels are the limit for one MIDI hardware link. So, as a composer, you'd assign all your tracks to the 16 channels. But if the resulting output exhausts the MIDI device's polyphony, another hypothetical 16 channels won't change that (again, the MIDI hardware link is limited to 16). So, by splitting the channels (odd/even), two MIDI devices with a given polyphony have less voices to generate (individually), and therefore, their individual polyphony limit has been effectively circumvented (as now two MIDI devices share the work intended for one). The line outs of both MIDI devices are then combined in my audio chain (with a 3u rack mixer).

I would like to have a setup like this. The challenge is finding an exact match for the existing SC-55, since there are so many different firmware revisions. Most sellers don't know how to display this information, and many units on auction don't even come with a power cable for them to test with.

I still don't quite get what the benefit of this is ^^ is there any realistic scenario (apart from complex music making) where you'd run into a limit without doing this?

imi wrote on 2022-04-27, 16:05:

I still don't quite get what the benefit of this is ^^ is there any realistic scenario (apart from complex music making) where you'd run into a limit without doing this?

Quite a few games have MIDI that occasionally surges past the polyphony of the classic MIDI modules that were used to create the music. Only the last of the premium MIDI hardware modules can fully play such music. That's why software MIDI is so great, as you can easily get 128-256 and beyond with modern hardware.

RetroGamer4Ever is correct. The Yamaha MU50 and Roland SC-88 are probably the best options for maximum polyphony using a single module. The SC-55 though has a very unique and, in my opinion, the best sounding instruments out of all the early General MIDI devices. But it has a number of other issues, including the aforementioned limited polyphony, which is present in games that, although era-appropriate, you might not have realized are clipping or dropping notes, such as in Warcraft II.

The SC-88 in SC-55 compatibility mode is the closest alternative using genuine vintage synthesizers, but it doesn't sound exactly the same. Even the SC-55 itself has a small range of differences depending on which hardware and firmware revision is used.

One interesting thing to note about Warcraft II is that the mastered soundtrack is actually a combination of the SC-55 and SC-88. I don't recall where I read this exactly; it has been quoted somewhere on this forum.

RetroGamer4Ever wrote on 2022-04-27, 16:09:

imi wrote on 2022-04-27, 16:05:

I still don't quite get what the benefit of this is ^^ is there any realistic scenario (apart from complex music making) where you'd run into a limit without doing this?

Quite a few games have MIDI that occasionally surges past the polyphony of the classic MIDI modules that were used to create the music. Only the last of the premium MIDI hardware modules can fully play such music. That's why software MIDI is so great, as you can easily get 128-256 and beyond with modern hardware.

do you have any examples by any chance? I'd be interested to hear the difference it makes.

imi wrote on 2022-04-27, 19:47:

do you have any examples by any chance? I'd be interested to hear the difference it makes.

See here:

Significant differences with certain MIDI files across Roland synthesizers
An analysis of Warcraft 2 MIDI music on Roland synthesizers

I put together a lot of very detailed samples there that demonstrate what it sounds like to run out of polyphony during playback. The artifacts are very subtle and will differ between synthesizers, firmware revisions, and sometimes even the same synthesizer when the same MIDI track is played in a loop. In practice, these variations normally go unnoticed, as any first-time listeners will have no other point of reference. The differences only become apparent when listening to the same track when played between synthesizers.

imi wrote on 2022-04-27, 19:47:

RetroGamer4Ever wrote on 2022-04-27, 16:09:

imi wrote on 2022-04-27, 16:05:

I still don't quite get what the benefit of this is ^^ is there any realistic scenario (apart from complex music making) where you'd run into a limit without doing this?

Quite a few games have MIDI that occasionally surges past the polyphony of the classic MIDI modules that were used to create the music. Only the last of the premium MIDI hardware modules can fully play such music. That's why software MIDI is so great, as you can easily get 128-256 and beyond with modern hardware.

do you have any examples by any chance? I'd be interested to hear the difference it makes.

Many of the Quest Studios (Sierra) MIDI files bump past the limit of the 90's MIDI hardware modules. I've seen them go into the 100s, on occasion. That's why the X2GS is a no-brain buy for a retro gaming build, because it can handle much of the old game music, with it's polyphony of 81. The next model from Dreamblaster will have even higher polyphony and handle anything from the golden days of MIDI gaming.