Sequence 01 (7.42s - 0.86s). Here the bell tone is shortened on the SC-55mkII and SC-55ST. It sounds fine on the SC-55 and SC-7.

Here is the same sequence after the voice reserve has been rearranged. This shows a clear improvement on the SC-55mkII but the SC-55ST does not benefit at all.

Sequence 02 (4.75s - 6.00s). Despite the improvement to the SC-55mkII had with the voice reserve SysEx, this part of the track remains unchanged. It sounds fine on the SC-55 and SC-7 but is short on the SC-55mkII and SC-55ST.

Sequence 03 (2.5s - 3.3s). The SC-55mkII shows no improvement even with the SysEx command. The bell is also cut short here on the SC-55ST, as well as on the SC-7. On the SC-55 I don't think the bell has been shortened at all.

Another observation which I found to be consistent across the SC-55, SC-55mkII, SC-55ST and SC-7 is that the strings instrument which plays in the background has an audible oscillation. I don't know if this is an artifact of the limited polyphony of these synthesizers or if it's related to the sample used in ROM, but it sounds strange. I would argue that, because the music for this game has issues with the polyphony on these older synthesizers, that this was not the author's intent.

To provide a contrast, here is the same 9 second sequence but performed by the SC-88 native as well as the SC-88 in SC-55 mode. Notice how the strings do not oscillate as they did with the other synthesizers.

The bell which was an issue for most of the other synthesizers sounds perfect here with a long and deep ring to it.

Here are recordings of the entire track from all of the synthesizers for comparison if anyone would like to try and find anything else to use for comparison.

The same track from the SC-55mkII but after the SysEx command has been applied.

For further comparison, here is the full track on the SC-88 in native and SC-55 mode.

The conclusion is that the music for Warcraft 2 needs at least a SC-88 to take full advantage of expected voice polyphony. This is confirmed by the older synthesizers which have issues reproducing the bells in the first few seconds of this track. The SC-55 makes a surprising show being able to produce a perceptively better reproduction than all of the polyphonically superior SC-55mkII and others. However, these recording are the best-case scenario where only one song is played with nothing preceding it. Even though the SC-55 seems to perform well in this example, it too can manifest the same shortened bell sound given the right circumstances. For example, the bell at the very start of the track will be cut short if the track is played on a loop rather than from the start in isolation.

I happen to favor the sound of the SC-55 over the SC-88 but there are still clear benefits to owning an SC-88 for situations where the former is unable to keep up. It's in my opinion then that the best reproduction for Warcraft 2 is with a SC-88 but using the SC-55 mode samples. While it won't sound exactly like an SC-55, all of the instruments are guaranteed to be present.

And just for fun, here is the same track as produced by a Yamaha MU50 for an alternative performance.

Is there a way to query what the voice reserve layout is on the SC-55 and configure the SC-55mkII to match it? I'm curious how it will sound in this configuration since it already sounds worse by default.

Manual method
Press both the < and > Part buttons simultaneously to access the settings menu for the currently selected Part, then press the Mute button seven times to scroll down to the Partial Reserve parameter. At this point, you can press the left and right Part buttons to check the setting for all sixteen parts.

SysEx method
Send and receive the following string:
F0 41 10 42 11 40 01 10 00 00 10 1F F7

The default response will be as follows:
F0 41 10 42 12 40 01 10 02 06 02 02 02 02 02 02 02 02 00 00 00 00 00 00 17 F7

Just to spoil the surprise, the SC-55 and SC-55mkII share the exact same default voice reserve settings.

The difference between the two relates to the voice assigner wall, which creates a division between parts, and prevents voice stealing from occurring between the separate groups (for voices that are already sounding). Despite the fact that the SC-55 and SC-55mII report the same value for the wall parameter, the default behavior of the SC-55 is a separation between parts 1-10 and parts 11-16. By contrast, the behavior with the SC-55mkII is one of no separation at all, meaning voice stealing can occur amongst all parts.

I've mentioned it before, but adjusting the voice reserve is an impractical solution - it gets reset to the default settings by reception of either a GS Reset or GM System On message, and therefore can't be applied to most game-related scenarios. The only real solutions are to either use a tone generator with greater polyphony, or, as specifically applies to the SC-55 and SC-55mkII, expand the polyphony by splitting the MIDI channel reception between multiple units.

Is there a guide on how to split the MIDI channels between multiple units? I would like to try this.

Kahenraz wrote on 2022-02-04, 18:05:

Is there a guide on how to split the MIDI channels between multiple units? I would like to try this.

I'm partial to hardware methods, and have one setup that involves the use of MOTU MTP units, and another with just a Music Quest MQX-32M interface card; either of which allow for such splitting.

Would you explain how this works with your setup, please?

can the SC55 do overflow via midi through? I know I can double the polyphony with my EMU SoundEngine by linking two together and one overflows to the next.

Roland SC-88 came with a 64-voice polyphony as well as the first Yamaha XG compatible synth, the MU-80. Although there were cost reduced models like MU-50 with only 32 voices. Although that's year 1994 for the SC-88 and 1995 for the MU-80 while the SC-55 was introduced in 1991. I recall E-mu Proteus 1 was launched even before the SC-55 and was capable of 32 voices, but it wasn't GM compatible for obvious reasons.

BloodyCactus wrote on 2022-02-04, 22:46:

can the SC55 do overflow via midi through? I know I can double the polyphony with my EMU SoundEngine by linking two together and one overflows to the next.

I found that there is a bit of lag between synthesizers when chaining them together via MIDI thru. If you chain a couple of them together, for example, the last in the chain will be very out of sync with the first. Can anyone else confirm?

I always preferred Warcraft 2 soundtrack on SC-88 (but then again I tend to favor 88 over 55 most of the time anyway even if I know it was a different model used to compose). Not sure if Glenn Stafford composed it on SC-88 but it definitely was some Sound Canvas.

According to a post on vogons by lordskylark referencing a quote from Stafford its both a 88 and a 55? Which might explain things

Warcraft II Music Soundcard

For Warcraft II -- from Glenn Stafford

"The original equipment used to record the Warcraft II music was a Roland SC-55 layered with a Roland SC-88. I recall only certain sounds were doubled with SC-88 but not all. French Horns were doubled, and... probably some of the woodwinds too, and I’m guessing strings and choir too. I think Snare was better on SC-55 alone, maybe timp too, also reverse cymbal, and not entirely sure about the rest and would have to take it case by case."

Kahenraz wrote on 2022-02-04, 22:24:

Would you explain how this works with your setup, please?

With the MTP, it's just a matter of routing a given input port to any combination of output ports, and then muting MIDI channels on the output ports as appropriate. E.g., input 1 routes to outputs 1 and 2, with the even MIDI channels muted on output 1, and the odd MIDI channels muted on output 2.

It's a similar case with the MQX-32M. That is, output 1 goes to SC-55 #1, and output 2 goes to SC-55 #2, with even/odd MIDI channel muting configured to take place between the two. The benefit here is that the "splitting" all takes place directly from the (intelligent-mode MPU-401 compatible) MIDI interface card itself, without the need for external hardware.

BloodyCactus wrote on 2022-02-04, 22:46:

can the SC55 do overflow via midi through?

Not by any known or documented means.

Kahenraz wrote on 2022-02-05, 06:41:

I found that there is a bit of lag between synthesizers when chaining them together via MIDI thru. If you chain a couple of them together, for example, the last in the chain will be very out of sync with the first. Can anyone else confirm?

I find it to be a non-issue. MIDI latency compounds by ~3 ms, or 0.003 seconds, for each successive device in a chain. Where latency is a concern, the general recommendation is to not exceed chains of more than three devices, wherewith the difference between the first and last device in the chain is an imperceptible ~6 ms.

AppleSauce wrote on 2022-02-05, 14:12:

According to a post on vogons by lordskylark referencing a quote from Stafford its both a 88 and a 55? Which might explain thing […]
Show full quote

According to a post on vogons by lordskylark referencing a quote from Stafford its both a 88 and a 55? Which might explain things

Warcraft II Music Soundcard

For Warcraft II -- from Glenn Stafford

"The original equipment used to record the Warcraft II music was a Roland SC-55 layered with a Roland SC-88. I recall only certain sounds were doubled with SC-88 but not all. French Horns were doubled, and... probably some of the woodwinds too, and I’m guessing strings and choir too. I think Snare was better on SC-55 alone, maybe timp too, also reverse cymbal, and not entirely sure about the rest and would have to take it case by case."

So, it seems like recreating the score in a DAW with today's wavetable sampling tech is definitely the best way to go for recording the music as it "should sound", which is like a real soundtrack/symphonic orchestra.

Just want to say that my setup pretty much mirrors your findings exactly. I have two SC-55s, one is the older version with the 1.20 control rom (only the GS logo on the front panel) and the second is the slightly newer version with the 2.00 control rom (has the GS and GM logos on the front panel, still only 24 voice polyphony) as well as an SC-88Pro.

Both SC-55s have the oscillating strings effect and the bell is also cut off at times. The SC-88Pro in native mode and in full SC-55 compatibilty mode play the track perfectly, without the oscillating strings and the bell isn't cut off at all. I tried muting channels on the SC-55 panel but it didn't seem to make a difference. I guess the muted channels are still playing, just at zero volume, so the polyphony is still affected.

I'm gonna build one of these to split the channels between the two SC-55s via hardware, pretty much doing the same thing as what Cloudschatze has done with his setup.

https://diyelectromusic.wordpress.com/2022/08 … -router-part-4/