Having used a Terratec Maestro 32/96 (which was using the same sample ROM as the SC-55), I can tell you with confidence that the crysis font doesn't sound even remotely accurate in games. Everything is completely wrong.

You can always compare recordings on youtube of a real SC-55 in various games, and then try these games yourself with the crysis font. It is COMPLETELY wrong in everything 😵

realnc wrote:

Having used a Terratec Maestro 32/96 (which was using the same sample ROM as the SC-55), I can tell you with confidence that the crysis font doesn't sound even remotely accurate in games. Everything is completely wrong.

You can always compare recordings on youtube of a real SC-55 in various games, and then try these games yourself with the crysis font. It is COMPLETELY wrong in everything 😵

Well, I haven't been using Crysis GM for quite a long time; my staple sound fonts are still SGM, Bellatrix, and Drums! by Slavo. Now, I don't know how do brass instruments in SC-55 sound font sound, but so far, the best brass instruments for classical/orchestra MIDI songs, like that of TIE Fighter and Wing Commander, are still Bellatrix Orchestra.

I have tried Roland Virtual Sound Canvas, Yamaha S-YXG50, and various sounds from small to large, and TIE Fighter opening song still sound ugly, almost FM-like. But after I installed Bellatrix Orchestra, the brass instruments become right. Wing Commander opening song sounds great as well.

How about SC-55 sound font? How do its brass instruments sound?

Kreshna Aryaguna Nurzaman wrote:

How about SC-55 sound font? How do its brass instruments sound?

Somewhat similar to how it's supposed to sound like. With a real SC-55 module, it sounds like this:

https://www.youtube.com/watch?v=G0qXEknmAY8

The 24MB soundfont I linked to sounds the least wrong compared to that. Still wrong, but everything else sounds more wrong.

realnc wrote:

Somewhat similar to how it's supposed to sound like. With a real SC-55 module, it sounds like this: […]
Show full quote

Kreshna Aryaguna Nurzaman wrote:

How about SC-55 sound font? How do its brass instruments sound?

Somewhat similar to how it's supposed to sound like. With a real SC-55 module, it sounds like this:

https://www.youtube.com/watch?v=G0qXEknmAY8

The 24MB soundfont I linked to sounds the least wrong compared to that. Still wrong, but everything else sounds more wrong.

Then I guess I don't like SC-55 brass instruments too much.

Too bad I don't have Bellatrix Orchestra recording myself, but my soundfont stack (SGM, Bellatrix Orchestra, and Drums! by Slavo) sounds pretty much like this (not mine) when playing TIE Fighter, which sounds better to my ears than the real SC-55 rendition you gave me. The real SC-55 brass instruments still sound "FM-like" somehow. Maybe it's what the composer intended (TIE Fighter soundtrack was composed using SC-55, wasn't it?), but it's less pleasing to the ears than the youtube video I linked above.

Kreshna Aryaguna Nurzaman wrote:

Too bad I don't have Bellatrix Orchestra recording myself, but my soundfont stack (SGM, Bellatrix Orchestra, and Drums! by Slavo) sounds pretty much like this

Yeah, that sounds horrible to me, because it's not how I remember it 🤣

realnc wrote:

Kreshna Aryaguna Nurzaman wrote:

Too bad I don't have Bellatrix Orchestra recording myself, but my soundfont stack (SGM, Bellatrix Orchestra, and Drums! by Slavo) sounds pretty much like this

Yeah, that sounds horrible to me, because it's not how I remember it 🤣

Ah, so you're like an audiophile: you want to preserve fidelity, to hear what the composer intended, don't you? 😉

Well I am an audiophile, but in regards to MIDI, I prefer better-sounding sound font even if it means compromising fidelity. Same goes for graphics. If DOSBOX interpolation modes look better than real CRT for low-res (320x200) games. 😀

TIE Fighter re-orchestrated sounds much better than SC-55 version, despite it doesn't sound the same with what the composer intended (because the composer uses SC-55, no?). Too bad, it's not GM sound font (or so I hear). 🙁

The SC-55 (and the Maestro 32/96, which is the one I had) were not on the "hi-fi" side of things. They don't sound that great. And how could they. These were cheap devices, at least compared to professional synths of the day that cost thousands of dollars. Not so cheap when viewed as computer peripherals, but cheap compared to pro gear.

But for me it's about getting the same audio today that I was getting back then, including all shortcomings and audio glitches. For example, I prefer the MT-32 version of Monkey Island compared to the CD audio version. The CD audio is superior, but I want that MT-32 sound, because that's how I remember it 😀 (And is the reason I think Munt is the greatest thing to happen in the emulation scene since DOSBox itself.) This is why I prefer this particular SC-55 soundfont to other soundfonts; even though they do sound superior compared to the flat and "tinny" Roland sound, they don't sound the way I remember 😀

realnc wrote:

The SC-55 (and the Maestro 32/96, which is the one I had) were not on the "hi-fi" side of things. They don't sound that great. And how could they. These were cheap devices, at least compared to professional synths of the day that cost thousands of dollars. Not so cheap when viewed as computer peripherals, but cheap compared to pro gear.

No, I mean "hi fidelity" in the context of "accurate reproduction of what the composer intended". If TIE Fighter's (and most games's) music is composed using SC-55, then SC-55 is the highest fidelity GM player for such games, because it will reproduce exactly what the video game music composer intended --despite SC-55 doesn't sound really great on the first place. 😀

Let say you have high fidelity system --let say, a Cyrus 8SE CD player, Pass Labs XA-60.5 integrated amplifier, and Yamaha NS-1000 studio monitors. Very hi-fidelity, aren't they? But then you play an audio CD with low quality mastering; then your expensive, hi-fidelity system will reveal all the ugliness of the recording. 😵

Okay, so you probably trade the Pass Labs amplifier for warmer-sounding Marantz, or the Yamaha NS-1000 for warmer-sounding Wharfedales. Then the poorly-mastered CD suddenly sound better, because the warm-sounding amplifier or speaker (or both) hides the ugliness. But of course, it defeats the purpose of hi-fidelity.

Now, when it goes to audio, I'm more a hi-fidelity person, preferring transparent-sounding equipments instead of pleasant-sounding ones, because most albums I listen to are adequately mastered (mostly 1970s funk jazz or modern chillout albums, but I have listened to excellently-mastered trance music as well, believe it or not). In this case, "hi-fidelity" and "pleasant-sounding" are not contradicting each other.

Games, however, is another thing. Lots of game music and sound effects are of low quality. This is especially true for old games, with low-resolution .WAV or .VOC sound files. In this case, using transparent system is an exercise of masochism! 🤣 Because your sound system will reveal all the ugliness of those low-quality, lowly-sampled digital audio of such old games. That's why I'm using Yamaha Cinema DSP, because those DSP reverbs --which are looked down by audiophiles, by the way-- help hiding the ugliness of old game sound.

realnc wrote:

But for me it's about getting the same audio today that I was getting back then, including all shortcomings and audio glitches. For example, I prefer the MT-32 version of Monkey Island compared to the CD audio version. The CD audio is superior, but I want that MT-32 sound, because that's how I remember it 😀 (And is the reason I think Munt is the greatest thing to happen in the emulation scene since DOSBox itself.) This is why I prefer this particular SC-55 soundfont to other soundfonts; even though they do sound superior compared to the flat and "tinny" Roland sound, they don't sound the way I remember 😀

Ah, so you are a hi-fidelity person indeed. 😀

Pardon the necro, but I just found this interesting post:

Guitslinger wrote:

I'm working on a project that utilizes one softsynth that is not a cpu hog, and two instances of SFZ with one soundfont 15mb in size, and the other 54mb. Both instances of SFZ are maxed out performance-wise. The quality is set to 72--the most demanding, and the mode is SF32, also the most demanding. With a P4 2.4ghz cpu, my maximum cpu usage is 30% at the lowest latency setting for my ASIO driver mode, 4 ms. By adding a couple of more instances of SFZ with added soundfonts my usage will probably jump up to 70% or higher.

There are clearly limits to how many instances of SFZ one can run in a project. Keep in mind the RAM required to run a soundfont using SFZ is 2.5 times the size of the font. If you try to use five 50mb soundfonts, 1.25 gb of RAM is required.

Interesting. Could anyone confirm or deny this? And how about SF2?

Interesting. Could anyone confirm or deny this? And how about SF2?

Contrary to standard SF2 soundfonts where the samples are always in uncompressed PCM format, in case of SFZ soundfonts the samples often use some kind of compression. In the format specification only ogg-vorbis is mentioned ( https://sfzformat.com/legacy/) but you can find many SFZ soundfonts that use e.g. FLAC compressed samples. Compressed samples must be decompressed by Bassmidi (or by other soundfont engine) before playing, so the memory usage compared to the size of the compressed samples can be much higher depending on the compression ratio. There is no such problem with SF2 soundfonts. In case of SF2 you can roughly estimate the memory usage by the SF2 file size (always lower than file size in case of real-time dynamic loading of samples or equal to file size in case of pre-loading all samples).