My guess would be no.
Namely, the 32 KHz limitation would only apply to the samples. It is LA Synthesis, so the subtractive synthesis-part is not limited by the Nyquist frequency of the 32 KHz digital part.

Scali wrote:

It is LA Synthesis...

Correct me if I'm wrong but Sound Canvas modules like SC-55 and SC-88 are PCM modules, they don't use LA Synthesis (LA was used in older synths like MT-32, D-50, etc.)

To answer the OP, I don't think you should be worried about this 32khz stuff. You don't get this kind of module to obtain crystal-clear highs from very high quality samples. Sound Canvas can't compete with modern VSTs or large soundfonts, but they are great at what they do - provide a general palette of instruments in a smaller-sized ROM.

MMaximus wrote:

Correct me if I'm wrong but Sound Canvas modules like SC-55 and SC-88 are PCM modules, they don't use LA Synthesis (LA was used in older synths like MT-32, D-50, etc.)

It's what Wikipedia says:
https://en.wikipedia.org/wiki/Roland_SC-55

And well, SC does sound like LA, don't you think? Very D-50-like. It certainly doesn't sound like full wavetable.

If you're worried about 32KHz then don't be playing N64 games or SNES games. Most first party titles on the N64 have the sample rate locked to a approximate of 32 KHz, the S-MP of the SNES outputs at 32 KHz.

Hell even a few GameCube and Wii titles have their streamed music sampled at 32 KHz. It's not really that noticeable there either.

There are models of SoundCanvas much later than the SC-55 and SC-88 that don't output at a rate of 32 KHz and even mix at precisions higher than 16-bit but they probably don't even have the same sound bank anymore I would guess...

I would recommend trying to get a SC-88pro though, the stock bank on that has much nicer sounds than the SC-55mkii. But this is subjective.

Scali wrote:

It's what Wikipedia says: https://en.wikipedia.org/wiki/Roland_SC-55 […]
Show full quote

MMaximus wrote:

Correct me if I'm wrong but Sound Canvas modules like SC-55 and SC-88 are PCM modules, they don't use LA Synthesis (LA was used in older synths like MT-32, D-50, etc.)

It's what Wikipedia says:
https://en.wikipedia.org/wiki/Roland_SC-55

And well, SC does sound like LA, don't you think? Very D-50-like. It certainly doesn't sound like full wavetable.

I'm no expert but I think the wikipedia article is wrong on this. I agree that they sound similar - probably because a lot of the samples from the sound canvas series seem to have been obtained by recording the output of LA synths like the MT32 or the D50. For example if you listen to the famous "Fantasia" patch on a SC-55 (patch #89) - it is quite similar to an MT-32 outputting the same sound, but not the same. IMO the MT-32 (or even better, the D50) sounds richer, the sound seems to "breathe" more. In the Sound Canvas it's "flatter" sounding, because there is no LA synthesis element - it is only PCM sample.

In the LA modules, the sound is comprised of two parts, a very short PCM sample for the attack portion of the sound, and then a LA synth part for the sustain portion. The technology made sense because memory was very costly in the '80s and synth manufacturers could only fit a very small ROM on their machines. With larger memory chips become more affordable in the '90s, manufacturers turned more and more to PCM, because they could fit larger samples in their modules. PCM is usually better at approximating real instruments, but it's more dull than LA for "synth" sounds.

I suggest watching this excellent demo of the D50 - at 1:24 there is a brief explanation of how LA synthesis works:

https://www.youtube.com/watch?v=-x1fKX-tzWg

Scali wrote:

It's what Wikipedia says:
https://en.wikipedia.org/wiki/Roland_SC-55

The Wikipedia page is misleading, and is probably based on a broad interpretation of Roland's own SC-55 advertisement literature, in which it is described that, "Roland's LA Synthesis and RS-PCM techniques are acknowledged as being among the leading sound generation techniques in the world. The Sound Canvas's newly developed sound source uses the best of both of these techniques, resulting in sounds with superb quality and realism."

Where LA Synthesis is commonly associated with the "PCM attack sample + digitally-generated waveform sustain" portion of its sound-generation methodology, this isn't what's being referred to with respect to the Sound Canvas, which lacks the waveform-generation capability entirely. Instead, the Sound Canvas simply combines PCM playback with two other prominent LA Synthesis features - TVA (Time Variant Amplifier) and TVF (Time Variant Filter) processing.

Well, I go by logic: The D-50 was cutting-edge in 1987. The Sound Canvas was released as a budget sound module 4 years later. Full wavetable synthesis wasn't common yet, certainly not in the budget market (problem with full wavetable is that it requires a lot of memory, which was expensive). So, given the state-of-the-art, and the pricerange of the Sound Canvas, it should be LA, not full wavetable.

Also, no need to explain LA synthesis to me. I've playing with synths since the late 80s. Please, try not to be condescending. It is insulting.

Scali wrote:

So, given the state-of-the-art, and the pricerange of the Sound Canvas, it should be LA, not full wavetable.

Well, it is full wavetable. Logic be damned. 😀

Cloudschatze wrote:

Scali wrote:

So, given the state-of-the-art, and the pricerange of the Sound Canvas, it should be LA, not full wavetable.

Well, it is full wavetable. Logic be damned. 😀

Proof is where?

Scali wrote:

Cloudschatze wrote:

Scali wrote:

So, given the state-of-the-art, and the pricerange of the Sound Canvas, it should be LA, not full wavetable.

Well, it is full wavetable. Logic be damned. 😀

Proof is where?

All known documentation for starters, including manuals, MIDI implementations, service notes, and so forth.

Cloudschatze wrote:

All known documentation for starters, including manuals, MIDI implementations, service notes, and so forth.

I looked through the manual here, didn't find it: http://www.manualslib.com/manual/695759/Rolan … nvas-Sc-55.html
Care to point out where it says that? Because that is what I asked you before.

I'd thought that the manual mentioned the sound source being RS-PCM at least, but apparently not (the marketing documentation does). From the service notes though, "The SC-55 consists of a custom IC (IC24) integrating a PCM sound source, reverb, chorus, TVF and TVA, and three pieces of 1 Mbyte wave memory (IC26 - 28)."

Scali wrote:

Well, I go by logic: The D-50 was cutting-edge in 1987. The Sound Canvas was released as a budget sound module 4 years later. Full wavetable synthesis wasn't common yet, certainly not in the budget market (problem with full wavetable is that it requires a lot of memory, which was expensive). So, given the state-of-the-art, and the pricerange of the Sound Canvas, it should be LA, not full wavetable.

Also, no need to explain LA synthesis to me. I've playing with synths since the late 80s. Please, try not to be condescending. It is insulting.

I am sorry you feel insulted by my post - it wasn't my intention. 😕

I'm not sure I understand the logic though - there were already "full wavetable" modules pre-dating the sound canvas : the U-110 in 1988. As you say, full wavetable synthesis wasn't common by 1991, but this is exactly what made the Sound Canvas ground-breaking - it offered a broader range of sounds at a lower price point.

MMaximus wrote:

As you say, full wavetable synthesis wasn't common by 1991...

I'm not sure where this reasoning is coming from. By 1988, nearly every synth manufacturer was producing sample-based synthesis products (of varying price), giving way to a platform that spawned a decade of ROMplers based on those same technologies. Taking into account the mentioned U-110, Korg's M1, or even E-MU's Proteus (1989), 2 - 4MB of sample waveforms wasn't an uncommon amount, even then.

With the SC-55 though, and as relates to the original thread question, there seems to be some fairly heavy sample compression, so while some of the base sample waveforms can also found in other Roland products, both earlier (U-110/220, D-70, CM-32P, SN-U110-xx, etc.), and later (JD/JV/XV, SR-JV80-xx, SRX, etc.), there's a pretty noticeable loss of fidelity with their use in at least the earlier Sound Canvas line.

I have some old recordings comparing the PCM (CM-32P) part of the CM-64 with the Sound Canvas SC-55mkII, if they're of any interest.

MMaximus wrote:

I'm not sure I understand the logic though - there were already "full wavetable" modules pre-dating the sound canvas : the U-110 in 1988. As you say, full wavetable synthesis wasn't common by 1991, but this is exactly what made the Sound Canvas ground-breaking - it offered a broader range of sounds at a lower price point.

The logic is simple: it either had to be a late, 'high-end' LA synth, or an early low-end full wavetable synth, given the time and price. I thought it was the former (also because it does a reasonable impression of an MT-32, which is LA), but apparently it is the latter.
Later SC models seem to sound considerably better. Mostly because of much larger roms (as I said, that was the problem. Technically even the MT-32 synth can be used as a pure rompler, but the samples are very short loops).

In which case you obviously DO run into the Nyquist frequency with everything you do, so it won't be able to produce anything over 16 KHz, to get back to the initial question.

Thanks for all the replies and info.
So for my use (to listen to game midis and other midis for hifi) I would be better off with a large soundfont than a sc-55.

I was just curious because there is one for sale nearby and someone said that even a expensive high quality soundfont would be 'vastly inferior' than a hardware module. But what they probably meant was creating a classic sound for certain games how they originally sounded. I don't mind if the sound is a bit different from what a game would have used to sound like but I want quality. So probably the soundfont's I have would be better than buying a module it seems because with the 32khz sampling rate and the compression the quality won't be a good as a modern soundfont.

Where a sound canvas is likely superior to a soundfont, on old MIDIs anyway, is instrument volume balance. Soundfonts, although they can have much higher quality sounds, are a crapshoot in this regard.

IMHO if you're listening to songs created for the sound canvas, especially GS stuff, get a sound canvas. You might want to get a later model though, at least SC-88. Their SC-55 modes sound similar enough plus the SC-55 Mk 1 only had 24 note polyphony, which is too low IMHO.

Ok thanks. Lots of the music I am listening to is GM or GS but I never heard it on the original sound canvas and I don't really mind if it sounds a bit different from how to original game sounded, I just want quality sounds. So probably soundfonts are the way to go, which I have already. I bought a good soundfont but only the GM version, they don't sell a GS upgrade though, only if you buy the GM version again. Anyway, I'm stacking other soundfonts to hopefully get GS instruments as well but I'm not exactly sure if I'm getting GS instruments by stacking GM ontop of GM/GS, anyway that's another thread.