The thing is that Arachno and SGM both have the best audio mixing out of most of these GM soundfonts. IE, its mixed so that the instruments roughly are the same volume as the original instruments in the default midi playback. That's one of the main reasons why those two soundfonts sound so great in particular. Other soundfont makers, for some reason, don't realize that there is no real standardization to the audio (midis aren't even limited to 0db so you can have clipping as well, if sounds get to loud), so, ideally, they should be mixing the samples to be similar in volume to the originals so that compatibility is the highest. BUT that's not what people do! At all! So you have a ton of ones that are really badly mixed (not naming names) and only a couple that are doing things "right". Of course with Virtualmidisynth and stuff you can change the volume etc. but that doesn't fix one of the samples just being overly loud or too quiet. You have to fix those things manually, and at that point I feel like the original maker of the soundfont has somewhat failed in doing what they were supposed to be, you know, doing. I'm so OCD I just a use a sf2 that just simply remasters the default soundbank. Same sounds, but sampled at a higher bitrate.
What's your sf2 of choice in particular? I should say though, the criticism as if they're just lazy or dumb is uncalled for. It is a chief priority and task of every single GM soundfont designer to balance the instrument volumes for compatibility... but well the problem is you can't. Because you're not also matching the timbre and effective envelope of the instruments, since that's where the unique sound of your instruments lives! So for two different samples that represent the same instrument, there exists no relative gain that will sound similarly loud in all the different music pieces. Because let's say one of your samples has a louder attack section, that's the bit at the start of the sample, relative to the sustain portion, compared to the other sample, then it will sound louder in music pieces which play very short notes. And then even if it's not louder, if it has more noise, more character, something distinctive happening, it can still poke out and seem louder. And then in the sustain portion things can also sound fatter or thinner while being perfectly volume balanced, even perceptually volume balanced with A-weighting or the like and any other gain corrections, and they will either obscure other instruments or not in the mix, depending on whether these instruments preferentially hit the same or constrasting frequencies, and whether these have distinctive attacks or other traits or not. Then if you balanced out the instrument at one note... because your timbre isn't the same as original instrument, it doesn't have the same exact composition of overtones, a gain set against a whole range of notes has a diffferent effect on different notes, unbalancing them against reference. There is further timbre induced unbalancing happening at different velocity, if you look at Fletcher Munson curves, your spectral perception is volume dependent. So you can't just set sample gains to match SC-88 and be done with it. And once you set gains such that they fit one music piece, you're likely to find that some other piece at the opposite end of genre spectrum isn't doing so hot. So these are genuinely impossible tasks. And then of course you have the human element that most SF2 contain one or another unintentional bug in the parameters, simply because that's a lot to handle. Of course some sound designs will lend themselves more towards matching a given classic GM device (and me personally i don't like Sound Canvas sound that much, i think there are other valuable targets) and others less, not for the lack of effort.
Could not agree more. People ask me all the time, what MIDI modules should I get etc, and the first thing I say is have you tried the virtual MIDI soundfonts available for free out there. If so and you want more then lets discuss your MIDI gaming or composing plans. Also outstanding explanation by the way. I might reference this video in the future James. ;D
I may eventually have to do a 'for idiots' series later on if people are liking my simple analogies and explanations that much. Cheers for pitching in. For those who don't know, Vanessa has an incredible MIDI setup!
Do people also have sort of... a home sound of sorts? I got a Yamaha keyboard in late 90s and i also used a Chaos 8MB soundbank on my AWE32. So these are basically still my go-to, now with S-YXG50 for gaming purposes and a Chaos 12M V20. Of course i used Chaos 12M V19 for many years on my Live as well, and i still have my Yamaha PSR530 too. Basically no matter what i try i have all these soundfonts i don't think they're going to replace these things for me. Though it helps that Yamaha is Yamaha and they for sure spent more sheer cash and effort than anyone else in the industry balancing their instruments. Even Roland is sloppier, though helped by the fact that they effectively set the standard. And even Chaos is sort of remarkably well crafted, it has flaws, but they're not too glaring.
Timbres of Heaven attempts to blow my speakers out every time! It's a good clear Soundfont when playing back but I can never get a decent capture of it for recording purposes without it distorting horribly. Arachno for me feels more in line with how 90s games should sound.
Found this soundfont called Matrix SoundFont, and it has been my favorite since. The guitar, synths, drums, and practically everything makes each song sound like a Genesis single, and it is beautiful.
@@getoffmylawn6692 I repair and refurbish old IBM and Lenovo Thinkpads, the oldest one (600X) I have goes back to 1999 and I think is the only one that I have with native DOS support (and a soundcard) for MS-DOS based Soundblaster emulation. There's an old Pentium III desktop in our work lab that I think is going to be junked soon and I may grab that if it's got ISA slots inside that I can just stick a sound card into, and then put DOS on it.
@@terrydaktyllus1320 I was watching some videos of abandoned hospitals last night and there were random old computers complete with CRT monitors just lying there. It was such a sad sight! Those older IBM models have a reputation for being indestructible, it makes sense that they're in high demand for refurbishment!
Thanks for this heartful review, James. Once again 😉
A pleasure having you here as always, Maxime. Keep up the excellent work!
100% agree. Excellent Sound Font! I use it almost exclusively.
The thing is that Arachno and SGM both have the best audio mixing out of most of these GM soundfonts. IE, its mixed so that the instruments roughly are the same volume as the original instruments in the default midi playback. That's one of the main reasons why those two soundfonts sound so great in particular. Other soundfont makers, for some reason, don't realize that there is no real standardization to the audio (midis aren't even limited to 0db so you can have clipping as well, if sounds get to loud), so, ideally, they should be mixing the samples to be similar in volume to the originals so that compatibility is the highest. BUT that's not what people do! At all! So you have a ton of ones that are really badly mixed (not naming names) and only a couple that are doing things "right".
Of course with Virtualmidisynth and stuff you can change the volume etc. but that doesn't fix one of the samples just being overly loud or too quiet. You have to fix those things manually, and at that point I feel like the original maker of the soundfont has somewhat failed in doing what they were supposed to be, you know, doing.
I'm so OCD I just a use a sf2 that just simply remasters the default soundbank. Same sounds, but sampled at a higher bitrate.
What's your sf2 of choice in particular?
I should say though, the criticism as if they're just lazy or dumb is uncalled for. It is a chief priority and task of every single GM soundfont designer to balance the instrument volumes for compatibility... but well the problem is you can't. Because you're not also matching the timbre and effective envelope of the instruments, since that's where the unique sound of your instruments lives! So for two different samples that represent the same instrument, there exists no relative gain that will sound similarly loud in all the different music pieces. Because let's say one of your samples has a louder attack section, that's the bit at the start of the sample, relative to the sustain portion, compared to the other sample, then it will sound louder in music pieces which play very short notes. And then even if it's not louder, if it has more noise, more character, something distinctive happening, it can still poke out and seem louder. And then in the sustain portion things can also sound fatter or thinner while being perfectly volume balanced, even perceptually volume balanced with A-weighting or the like and any other gain corrections, and they will either obscure other instruments or not in the mix, depending on whether these instruments preferentially hit the same or constrasting frequencies, and whether these have distinctive attacks or other traits or not. Then if you balanced out the instrument at one note... because your timbre isn't the same as original instrument, it doesn't have the same exact composition of overtones, a gain set against a whole range of notes has a diffferent effect on different notes, unbalancing them against reference. There is further timbre induced unbalancing happening at different velocity, if you look at Fletcher Munson curves, your spectral perception is volume dependent.
So you can't just set sample gains to match SC-88 and be done with it. And once you set gains such that they fit one music piece, you're likely to find that some other piece at the opposite end of genre spectrum isn't doing so hot. So these are genuinely impossible tasks.
And then of course you have the human element that most SF2 contain one or another unintentional bug in the parameters, simply because that's a lot to handle.
Of course some sound designs will lend themselves more towards matching a given classic GM device (and me personally i don't like Sound Canvas sound that much, i think there are other valuable targets) and others less, not for the lack of effort.
Could not agree more. People ask me all the time, what MIDI modules should I get etc, and the first thing I say is have you tried the virtual MIDI soundfonts available for free out there. If so and you want more then lets discuss your MIDI gaming or composing plans.
Also outstanding explanation by the way. I might reference this video in the future James. ;D
I may eventually have to do a 'for idiots' series later on if people are liking my simple analogies and explanations that much. Cheers for pitching in. For those who don't know, Vanessa has an incredible MIDI setup!
@@getoffmylawn6692 IS this really the best all arounder there is?
Do people also have sort of... a home sound of sorts? I got a Yamaha keyboard in late 90s and i also used a Chaos 8MB soundbank on my AWE32. So these are basically still my go-to, now with S-YXG50 for gaming purposes and a Chaos 12M V20. Of course i used Chaos 12M V19 for many years on my Live as well, and i still have my Yamaha PSR530 too. Basically no matter what i try i have all these soundfonts i don't think they're going to replace these things for me.
Though it helps that Yamaha is Yamaha and they for sure spent more sheer cash and effort than anyone else in the industry balancing their instruments. Even Roland is sloppier, though helped by the fact that they effectively set the standard.
And even Chaos is sort of remarkably well crafted, it has flaws, but they're not too glaring.
Have you tried Timbres Of Heaven? I think it works great with Tie FIghter.
Timbres of Heaven attempts to blow my speakers out every time! It's a good clear Soundfont when playing back but I can never get a decent capture of it for recording purposes without it distorting horribly. Arachno for me feels more in line with how 90s games should sound.
@@getoffmylawn6692 Timbres is great for some games, but I always end up defaulting back to SGM or Arachno when using virtual MIDI.
Found this soundfont called Matrix SoundFont, and it has been my favorite since. The guitar, synths, drums, and practically everything makes each song sound like a Genesis single, and it is beautiful.
So are you using these soundfonts on real vintage hardware, or in emulators?
I'm using emulators exclusively as all my old hardware is long gone. Oldest computer I have that's still functional is from 2001.
@@getoffmylawn6692 I repair and refurbish old IBM and Lenovo Thinkpads, the oldest one (600X) I have goes back to 1999 and I think is the only one that I have with native DOS support (and a soundcard) for MS-DOS based Soundblaster emulation.
There's an old Pentium III desktop in our work lab that I think is going to be junked soon and I may grab that if it's got ISA slots inside that I can just stick a sound card into, and then put DOS on it.
@@terrydaktyllus1320 I was watching some videos of abandoned hospitals last night and there were random old computers complete with CRT monitors just lying there. It was such a sad sight!
Those older IBM models have a reputation for being indestructible, it makes sense that they're in high demand for refurbishment!
are you using fl studio?
@1:30
Thai fighter
Personally, I prefer Masterpiece of Arachno any day..