It’s a pretty obvious statement that, when it comes to streaming and writing about old video games, some vintage consoles definitely are more popular than others. It’s also true that, when it comes to the chip tuning scene, some hardware platforms are inevitably more popular than others. Speaking as both musician and retrogamer, what really interests me is that when you take them together, the consoles which are the most highly regarded among players don’t inevitably have the liveliest and most active chip scenes.
Take the Super Nintendo, for example. As a gaming titan it shifted upwards of 49 million consoles and acted as the petri dish from which a number of Nintendo’s best loved characters and franchises were spawned. Consequently, you’d expect it to be among the elite of the chip-tuned consoles, right? After all, If you head over to OC Remix It’s not as if there’s a shortage of iconic SNES soundtrack remixes.
Weirdly, however, the SNES’ audio capabilities don’t seem to be that widely discussed or exploited. True, they might briefly be raised in a playground-style “What console is better?” thread over at Neogaf, but that’s generally it. You’re definitely not going to find the small mountain of compositions that have been produced for hardware in the Game Boy and C64. What gives?
It’s a really interesting one. If we look back to it’s 90’s prime, the SNES’ audio capabilities were most definitely a big deal. From the comparatively perfect (though now hilariously tinny) rendition of famous compositions like the Star Wars theme, down to the fawning coverage of the (faux) surround-sound capabilities utilised by the likes of King Arthur’s World, sound on the SNES was always pretty important.
So why don’t we hear more about it today then? An important reason might very well be the architecture of the chip itself, so we should probably start by having a look there.
Interestingly, the three main 16-bit consoles were all very different in their approaches to sound generation and, as the latest of the trio, the SNES was in the best position to learn from the mistakes made by the opposition. We’re getting ahead of ourselves though – we should have a quick look at both the previous generation and the environment the SNES was born into before we look at the console itself.
Sound on the 8-bit consoles, then, was a relatively simple affair. Though its worth remembering that capabilities were slightly different from machine to machine (there’s a very good reason that the C64’s SID sound chip is so popular,) they all utilised relatively simple Programmable Sound generators (PSGs.)
Generally, these PSGs featured 3-4 sound channels (Think of these as the number of instruments in the band) and these channels could produce any one of a small number of simple sound wave forms (think of these as types of instrument that are being played.) This was also often accompanied by the further limitation of dedicating one entire channel to white noise production (that’s static to you and me – Great for cymbals and explosions.)
Though these stark limitations lead to some pretty amazing pieces of music being created by a number of innovative musicians, they were limitations nonetheless – and it was obvious that the next generation of consoles would require sound hardware with a bit more oomph. Indeed, not only had Sega experimented with a different sound chip for the Japanese version of the Master System, but at the same time Konami were creating NES cartridges that contained their own sound architecture and Nintendo themselves were dabbling with a different sound solution for their Famicom Computer Disk System. As the eighties progressed into their final years, video game sound hardware was clearly on the move.
Though NEC’s PC Engine was the first out of the blocks in terms of release, when it came to the design Sega’s Megadrive was arguably the oldest of the trio. You see, Just as the rest of the Megadrive hardware was based quite firmly on the capabilities of Sega’s System-16 arcade boards (a smart move considering the potential for ports!) when it came to sound Sega also relied on the same technique they’d used in all of their arcade machines from the mid eighties onwards: FM Synthesis.
Intermission! Buy my stuff!
Was Sega’s approach a good idea? Well, from the perspective of building on the capabilities of earlier machines, FM synthesis definitely wasn’t a bad choice. Where earlier PSGs had 4 channels that could produce a handful of different waveforms, the Megadrive’s YM2612 had a whopping six channels that could produce four separate instances of one solitary wave form: a sine wave.
If this sounds like a step backwards in some ways, it wasn’t. By layering these four sine waves on top of each other – just as sine waves are in the sounds you hear around you every day – the YM2612 could not only recreate most of the basic wave forms generated by the earlier PSGs, but it could go beyond them to create a more diverse palette of more realistic instrument sounds.
Of course, there were downsides. Though more realistic than what had come before, an FM synth will never convince you that its the real deal for most natural sounds (for reference, FM synthesis is essentially the defining sound of 80s pop music – Think of all of those glorious Brass sounds.) It’s also worth pointing out that the Megadrive’s FM channels could generate just two thirds of the sine waves utilised by Yamaha’s more expensive synthesisers, giving it a slightly less refined quality. Still, on paper there’s no denying that it was a pretty formidable bit of kit.
However, while FM synthesis was good enough for the likes of Sega and Konami, NEC and Hudson took an altogether different approach for their PC Engine. Rather than synthesise sounds from scratch, the PC engine made use of a concept called Wavetable-lookup synthesis. This means that, effectively, the soundtracks to all of your favourite Turbo Grafx games were created using recordings of digital audio.
Wondering how they could fit so much music on such a small cart? Well the important thing to remember is the sample recordings were of the base instrument sounds – not the actual songs they were playing. This means that, if you had a passage containing thirty notes, the cartridge would only need to contain a short sample of one solitary tone. This could then be sped up and slowed down slightly to reproduce the rest of the required notes by the melody. Clever, yes?
Even if they were short, there were also some further (and more draconian) size limitations placed on these tiny instrument samples however. Not only did they have to be rendered at a 5-bit resolution (remember, 16-bit amounts to CD quality sound) but they could also be no more than 32 bytes (not KB) in size. Blimey.
That’s not to say that the HuC6280 chip was bad in anyway. By having access to 6 channels capable of loading these arbitrary wave forms, PC Engine composers had access to a larger number of channels and more diverse array of sounds than composers had when working with the earlier NES and Master System PSGs, even if the resulting music always had that characteristic angry wasp-style low resolution buzz.
The battle lines were drawn, then. Were Nintendo going to go down the NEC/Hudson Soft’s wavetable-lookup route, or follow Sega’s attempt to bring arcade FM technology into the living room?
As we saw earlier, the late release of the SNES left Nintendo perfectly placed to target any perceived weaknesses in the opposition’s hardware. Just as they sought to develop a console that could display a much wider array of colours than the Megadrive, Nintendo made a smart decision to partner up with a specialist audio hardware company in order to ensure that their Super Nintendo had the strongest possible audio capabilities.
Presumably as Sega were so closely linked to FM-patent holders Yahama, Nintendo opted to work with another Japanese giant who had a solid grounding in both audio and electronics: Sony.
And that’s not just Sony in general either – the sound unit for the SNES was to be created by a certain young engineer at Sony called Ken Kutaragi. He’d later go on to some slightly more ambitious projects.
So then, what route did Kutaragi and Nintendo take? Well, as the 1980s made way for the 1990s, big changes were afoot. This was particularly true in the digital instrument market, where the sudden crash of memory prices had made it just as economical to stuff a basic consumer keyboards with samples of real instruments than to try to emulate them with budget synthesiser chips. Though FM synthesis would live on as an interesting creative tool, it was becoming clear that the future of music creation would be digitally sampled instruments.
Consequently, it’s probably unsurprising that Nintendo opted to follow the path that themselves had helped establish with the Famicom Disk System, and go down a wavetable synthesis-based route. What is surprising, however, is just how far above and beyond the call of duty Kutaragi decided to go.
You see, where the Megadrive and the PC engine had sound chips, the SNES had an entire audio subsystem – the Nintendo S-SMP. This impressive unit was made up of its own 8-bit processor (which had to be programmed separately from the main system) a digital signal processor and its own (64kb) pool of ram.
Why was the unit Kuturagi created so much more complicated than the competition? Well, while the theory behind the sound units of the SNES and the PC Engine were comparable, it’s worth noting a couple of further short comings in the competition.
Not only was the PC Engine’s HuC6280 limited to the tiny and extremely low-res instrument wave samples we discussed earlier, but if composers on both the PC Engine and the Megadrive wanted to include longer stretches of sampled voice and sound effects, their only choice was to shut off one of the main audio channels and deploy unwieldy uncompressed 8-bit samples.
Unfortunately, these samples were relatively large and, to add insult to injury, they also meant much of the heavy lifting had to be done by software-based sound drivers. This approach not only had potential side effects on other areas of the game but, as this project to improve the quality of the voice samples in Megadrive Street Fighter II demonstrates, often resulted in questionable audio quality.
The SNES’ S-SMP, on the other hand, was a complete sound solution that addressed all of these issues. By building the sound chip as a self-contained co-processor, Kuturagi not only ensured that all of the complicated audio work could be handled without troubling the SNES’ relatively weak processor, but it allowed him to work with a compressed 9-bit audio format. This meant that, while the PC Engine’s wave table samples were limited to incredibly short fragments or sound, the SNES was capable of working with the longer stretches of audio you would expect to find used by a bonafide digital sampler.
Though it’s worth noting that the 64kb ram pool was never optimum and always somewhat of a compromise, as the composer still had 8 sample channels to work with and they could deploy the ram as they saw fit, the SMP still provided sound designers with a lot of options. It didn’t matter if the sample they wanted to load on a particular channel was a fragment of a sampled trumpet they intended for use as part of the background music or a recording of Darth Vader’s breathing intended for use as a sound effect – Kuturagi’s design effectively offered games designers a complete and relatively high-res solution to all of their audio needs.
Indeed, throw in the SNES’ trademark digital echo effect and a bunch of helpful filters/automation tools, and the end result was a chip that could easily go toe-to-toe with the Megadrive’s YM2612.
Just to illustrate that last section, here’s a few instrument samples which have come straight from the cartridge of Super Star Wars. Note that the brass sound isn’t even a complete sample of the note – The sustained notes in the theme were actually created by infinitely looping a very short section of audio:
That’s not to say Kutaragi’s solution was perfect, mind you. As the S-Smp could only load 64kb’s worth of audio data at any one time, one massive oversight on Kutaragi’s part was an inability to load new audio samples on the fly at a hardware level. By design, the S-SMP was supposed to grab all of the required audio data at start up, and that was that.
Thankfully this limitation was soon worked around (most notably by Streets of Rage composer Yuzo Koshiro,) but it remains an important oversight in the design that’s worth mentioning.
Overall then, it’s difficult to argue against the notion that the design of the SNES audio hardware was a more modern and forward-thinking design than the technology deployed in the Megadrive and the TG-16. We have to wonder then, why does it lack a large dedicated scene today?
Sadly for the SNES, it’s largest advantages back in 1991 are arguably its biggest achilles heel today. While the YM2612 remains both relatively well documented and easy to program, up until recently the S-SMP had remained a relatively inaccessible enigma.
Not only was the technical documentation for it distinctly thin on the ground, but from an assembly perspective its self-contained nature meant that it had to be handled differently from the rest of the SNES. These factors combined so that, historically, there simply haven’t been the tools available for musicians to make their own SNES music.
On top of that, when working with the SNES, there’s is an inevitable existential question of just why you need to use SNES samples in the first place.
The SNES’ position in the history of game audio is a complicated one. While, on the one one hand it lacks the highly distinct lo-fi character you get from the Metallic twang of the YM2612 or the aggressive low-resolution buzz of the HuC6280A, on the other the sound it produces isn’t of high enough quality for it to be a fun replacement for a modern sampler.
In that sense it feels very much more like an obvious stop-gap than either a basic PSG or the Megadrive’s FM Solution. You get the feeling that Kutaragi knew that the CD-quality soundtracks we enjoy today were an obvious inevitability, but they were just a little bit too far out of reach to be recreated by the hardware he had access to (well, without some sort of add-on, anyway…)
This, unfortunately, means that SNES samples are a lot harder to contextualise when they’re used outside of a SNES game – Is that a SNES you’re hearing or just a really cheap General Midi keyboard? Is it supposed to sound like that? Depending on the composition, it can be hard to make clear that the SNES is exactly the sound you’re going for.
So then with all that in mind, is SNES music something that’s best enjoyed as a listener rather than a creative? Absolutely not! Once again, it all boils down to the architectural differences between the SNES and the other 16-bit machines. Though it’s true that back in the 90’s the sound chip’s biggest selling point was it’s ability to playback realistic samples, it’s actually capable of doing so much more.
Indeed, While the Megadrive and PC Engine created their character in how they generated sound, as a creative tool the SNES’ strength arguably lies in how it can modulate and process sound. Along with a pretty distinctive echo effect, the SNES is capable of doing lots of clever tricks around envelope generation and pitch modulation (an envelope is basically a recorded setting for adjusting an instrument’s parameters over a period of time.)
This means, for example, that the SNES can take one channel of sound, adjust its parameters in a way dictated by an envelope, and then use it to effect the pitch of a separate channel. This allows you to do lots of awesome stuff – especially when it comes to warping instrument sounds from well known and well loved games. Though the SNES can sound a bit like a cheap MIDI keyboard when used out of context, the truth is it doesn’t inevitably have to.
So there you have it. 25 years after its original US release, the SNES’ sound chip is still posing some interesting questions. Personally, I wouldn’t say the S-SMP is my favourite sound chip, but it definitely is an interesting bit of hardware that occupied an important spot in the development of video game music. I hope that, when you next fire up your favourite SNES classic ,you take a short moment to really try and hear the hardware behind the music.
Oh, you’re still here? Well, if you’re interested in making use of the sounds of the S-SMP yourself, you have a few options available to you these days:
For those who want to write music for an actual SNES, up until relatively recently you’d have been out of luck. However, back in 2014 A Russian developer called Shiru created something dubbed the SNES GSS (Game Sound System) and a pair of developers called Ferris and cTrix have also been busy putting together their own independent tool set. Neither of these are the finished article and, as tracker programs, they work a bit differently to more common digital audio workstations. Still, for those who inhabit the venn diagram between progamming and musical proficiency it’s at least nice to have the option.
For those less interested in the technical features of the SMC chip and more interested in converting particular SNES sounds into something they can use in a modern sampler, don’t worry – you’re in luck as well! A helpful chap called Bregalad has created a toolset capable of ripping the SNES’ compressed BRR format into uncompressed (and widely usable) WAV files. You can even go the other way, if you really want to. Hurrah,
Perhaps the most interesting option, however, is a VST instrument created by Osoumen over at picopicose.com . His C700 is effectively a fully functional S-SMP emulator that can be run in any VST/AU-compatible Digital Audio Workstation. Along with being able to grab samples straight from .SMC music rips, the C700 is even able to convert external wav files to SNES BRRs on the fly. Nifty. Unfortunately the manual is in Japanese, and I haven’t had much time to test it, but with the SNES’ iconic echo sound and envelope controls in place (along with the ability to go beyond the limitations of the original hardware) mean this is an option that’s definitely worth investigating.
This article was updated on the 22/01/2018 to improve its clarity in a couple of areas.