Modern games consoles, eh? With their bundled hard drives and insatiable hunger for game patches, they’re effectively just user-friendly PCs aren’t they? Yes, even you over there in the corner, Mr Switch.
Hang on though! When you think about it, the inverse is also true. While console games have grown increasingly complicated since the days of plugging chunky plastic cartridges into a hidden slot, It’s now no more complicated to update a PC game on Steam than it is on an Xbox or Playstation – especially now Microsoft have decided to get in on the act with their own version of Apple’s app store.
Mind you that may not be a bad thing. Back in the DOS era, installing a game meant filling in a multiple choice quiz about your machine. Don’t think you could get away with simply knowing that your PC contained a Sound Blaster expansion card either. If you weren’t aware of the settings it used to talk to your processor the whole system could lock up. Aieee.
On top of that, just look at the choices! by 1991, it wouldn’t be surprising if a game had 9 different sound options. Why were there so many? Were they all different products that effectively made the same sound, or were they something different? Let’s go through them!
Debuting with the IBM’s original 5150 in 1981, the PC speaker was the first and most basic piece of sound hardware available to PC gamers. Though PC speakers have differed in design across years and manufacturers (early PCs used magnet-driven PC speakers, later models deployed piezoelectric designs), the theory was (and is) the same: the PC Speaker is a simple speaker that is built directly into a PC’s motherboard in order to offer basic feedback while going through the boot process. That reassuring chirp you still hear every time you turn on a windows-based PC? Yep, that’s the PC speaker!
The PC speaker is, arguably, both the most and least interesting of all the PC hardware. In the minus column, it doesn’t really have much going on in terms of features – in fact it doesn’t really have any at all. Not only does it lack the separate channels you needed to make different instrument sounds, it only really has the ability to create one type of tone (a square.) In fact, Aside from the physical switch controlled by the user, it doesn’t even have any direct contro over its volume either.
Essentially, then, the computer tells it to beep, and it beeps. That’s literally all it does. If you’ve never had the good fortune to hear one, think of any old mobile phone ringtone and simplify it.
With that said, the PC speaker did have a couple of things going for it. The first was its ubiquity: As it was built into every PC, it was a dependable standard from that was available from the earliest days of DOS-based PCs. There was no real excuse for not supporting it.
The second was that interaction between the speaker and the computer. If you were triggering it manually by pressing a note on a keyboard, the PC speaker would only be able to make a simple audible beeping noise. The beauty of the computer, however, is that it doesn’t have to interact with the speaker as you or I would. Instead, the computer can trigger and cancel the speaker multiple times a second. One immediate use of this was polyphony. By combining rapid pulses of different lengths, composers could give the impression of different instrument sounds and – by timing the off beats very carefully – even of instruments playing at the same time, as you can hear in the PC speaker version of Lucas Arts Monkey Island theme.
These rapid pulses weren’t just good for faking polyphony, by carefully generating and cancelling thousands of tiny pulses at different volume levels the PC speaker, the natural filtering properties of the PC Speaker allow it to be used as a primitive DAC (digital to audio converter.) This meant that the PC speaker can be used to play (very) crude versions of digital sound recordings.
Now before anyone gets too excited, crude in this instance really does mean crude. Sample playback via the PC speaker meant playing off the effective sample rate (the amount of times the converter references the original wave) against the effective bit depth (the amount of information it records every time it looks), and even if this was optimised, the resulting 18khz sample rate and 6-bit depth were well below even what you would expect from an audio cassette.
On top of that, low quality samples came at a large cost too. Not only did they take up (what was in this era very) valuable disk space, but all of the heavy lifting was performed directly by the computer’s CPU – meaning their were limits on the complexity of the audio work the computer could perform while simultaneously running game code. Unfortunately, By the time PC CPUs had enough grunt to overcome this performance bottleneck, more advanced pieces of sound hardware was becoming common place.
Still, despite all of that, PC speaker samples did manage to find their way into a number of commercial titles. The PC port of Space Harrier, for example, includes an impressive digitized scream while the final code for the mighty Wolfenstein 3D actually contains a fully functional PC speaker sound effect driver that removed before the final release.
The PC speaker is a funny old thing then. Though most of the soundtracks produced for it were built entirely from plain, monophonic beeps, in the right hands it had abilities that surpassed its meagre specs. The oldest of PC sound options, the PC speaker also happens to be the longest lived: it may have been baked into the earliest PCs, but the PC speaker will still provide its comforting chirp when you boot a brand new PC today – and it can even be used as a emergency output for Skype if your modern sound hardware fails. Blimey!
Considering the length of the PC speaker’s dominion of the PC sound market, it might be strange to learn that a superior option showed up as early as March 1984. Armed with a 3 channel sound chip, IBM’s PCJr should naturally have set a new standard in PC sound. In the event, things didn’t quite work out that way. Why?
Well, to summarise a slightly long and off topic subject, by 1983 IBM had effectively achieved dominion over the business sphere so decided to build a machine that would allow them to compete in the home as well. Though the machine they came up with – the PCjr – merrily boasted a number of useful home-centric features (such as the ability to connect to existing televisions and stereo systems), IBM hadn’t really focused on where the machine would fit in the home market. Priced at $1269, the jrs specs weren’t impressive enough to compete with premium machines like the Apple Macintosh on performance, while at the same time its price was too high to compete with relatively inexpensive micro computers on cost.
Consequently, though it launched in with tremendous media fanfare in 1984, reception to the machine was lukewarm. Sales inevitably fell well short of IBM’s lofty expectations and in the following year the firm decided to pull the plug. Had it not been for the Tandy brand, that would probably would have been the end of that. Instead, the platform got a new lease of life in the form of the Tandy-1000 clone. Sharing the base spec of the jr but fixing a few of IBMs more questionable business decisions and selling for a cheaper price, the Tandy-1000 may not have sold well enough to set a new universal standard, but it managed to make enough of a splash to garner support from a number of important PC game developers.
Was it a shame the the PCjr/Tandy soundchip didn’t manage to fully replace the PC Speaker? Well, the jr’s soundchip was definitely a step up from the PC Speaker in most respects. Equipped with both 3 square-wave-pushing tone generators and a white noise creator, the SN76489 used in the Jr/Tandy was capable of playing three different notes through three different instrument sounds with an extra channel left over for percussive effects. Because the SN chip was also a programmable sound generator instead of a simple speaker, it was also a lot more self sufficient than the PC Speaker and placed much smaller demands on the resources of the CPU.
In terms of sound, the chip was basic by contemporary standards but effective enough as a sound generator. As the only melodic tones it could generate were variations on the basic square wave, to the untrained ear its music may be indistinct from chiptunes created for the NES or C64. However, on closer inspection, though the SN chip lacks some of the features found on those systems, it goes some way towards making up for this with distinct, pure-sounding tones. It also allowed composers to create pieces that were more advanced than the ones they could create for the PC speaker, as when they weren’t used for polyphony, the tone generators could double up to produce rudimentary effects like chorus and delay.
In fact, the only real negative mark against the Jr sound chip is its relative lack of ambition. You may not have heard it used in the context of a PC, but if you’re enjoy retrogaming you’ll most likely have heard the SN76489 elsewhere: The chip found its way into the Sega SG1000, Master System, Megadrive and Pico, along with the ColecoVision, BBC Micro and Neo Geo Pocket; while the broadly similar AY 89-10/12 found its way into the Zx Spectrum 128 and the MSX standard. Considering that both Amiga and Apple were experimenting with solutions based around sampled audio, a common off-the-shelf solution was never really going to cut it in a product due to be released at a premium price point.
Overall then the Tandy/PC Jr soundchip wasn’t such a bad unit. It may not have shifted enough units to become the new benchmark in PC sound, but it found significant support from the likes of Electronic Arts, Sierra and Lucas Arts and new games still supported it seven years after its introduction. If you’re a fan of the chiptune sound, it’s definitely worth seeking out the SN764789’s lo-fi versions of some of DOS gaming’s defining soundtracks. Mind you, with all of that said, its’ worth noting the SN764789 was also never supposed to be the final word when it came to PCjr sound, as audio was one area that IBM specifically designed for the jr to be expanded. Though the option for audio expansion was there, It was to be a further three years before someone would design a sound card that would really take the market by storm.
AdLib Music Synthesizer
Considering that the PC speaker was effectively outmoded at launch, it still seems surprising that it took so long for a superior benchmark to really take hold. Nonetheless, three years after the launch of the PCjr, most PC gamers would be playing with PC speaker sound. Thankfully, 1987 saw the launch of the sound card that would change everything – the AdLib music synthesizer card. It didn’t take hold over night – there were still a number of titles released in 1989 that lacked AdLib support – but by 1990 it was essentially the default sound option for most games. If you can conjure a tune from an early ’90s DOS game into your head, chances are you’re probably thinking of the AdLib version.
The card was the brainchild of Martin Prevel, a Canadian music professor who had an interest in the potential computers held in the field of music. He had previously developed a specialist 8-bit computer designed to help train the human ear, and even though that had been a relative failure, he decided to have another go with a sound expansion card instead. The key field he identified was cost: After the failure of the Pcjr, IBM themselves had also attempted to market a sound card – the Music Feature Card – which was built around the same Yamaha sound chip used by a number of contemporary arcade machines. Unfortunately for IBM, they had another flop on their hands: retailing for half the cost of a PC at $600, the feature card barely sold 10,000 units.
Contacting Yamaha, Prevel stumbled on a fortunate happenstance: just as he was looking for a cheap soundchip to power his low-end expansion card, Yamaha had found themselves with a surplus of simplified FM sound chips after the cancellation of a Japanese government project. Though initially reluctant to export to Prevel, Yamaha eventually relented and the AdLib and music synthesizer card was born. Quickly utilised by Sierra Online (who were both enthusiastic early adopters and active resellers of PC sound hardware), other PC developers were quick to jump on the band wagon, eventually leading to AdLib becoming the cornerstone of DOS-based music.
What made the AdLib card so appealing? Well, aside from costing a third of the price of IBM’s $600 effort, the AdLib was a huge step forward in performance when compared to the PC speaker and Jr sound. Where the earlier sound options had a maximum of one tone generator per instrument and a maximum of three instruments, the AdLib card was capable of playing a whopping 8 different instrument sounds at once, with each sound being created by two separate sine wave generators.
This use of multiple sine waves to create a single instrument was a massive deal. The chip powering the Adlib card (the YM3812/OPL2) worked on the principals of FM synthesis. Though this can be a little complex to explain, it’s a bit like boiling a liquid: when you modulate one sound with another lower frequency sound you end up with a sound that’s akin to a low boil: You can hear the rise and fall of the lower-frequency modulator inside the original sound, creating an effect that sounds a bit like a siren. If you speed up the low frequency sound a bit more, you end up with the sound equivalent of a fast boil: the individual rises and falls of the modulator start merging together, causing the original sound to wobble.
This is all well and good, but what’s really interesting is what happens when the speed of the the lower frequency sound begins to match the frequency of the original sound. At this point the conjoined sound waves split into harmonically-linked duplicates that merge together. like boiling water turning to steam, this creates a new sound with very different properties to the original. Depending on the amount of sound generators you have at your disposal, you can use this technique to make an infinite array of new and interesting sounds.
For composers this was an exciting concept. With earlier hardware, the composer was forced to infer timbre through their arrangement – using, say, a stabbed low-pitch note to represent a picked bass guitar. However, thanks to the complicated sounds you could create with FM snythesis, it was now possible to create a timbre that sounded recognisably, though not indistinguishably, like a real bass guitar. Perhaps unsurprisingly then, FM Synthesis had a sudden and dramatic impact on arcade and computer music: Yamaha’s pioneering DX7 became one of the best-selling and most ubiquitous synthesisers of all time, while the release of the FM-equipped Marble Madness arcade machine drew a clear line in the sand between the 8 and 16-bit arcade eras.
Inevitably of course there were some drawbacks. After all, if both cards were built around FM synthesis, why was the AdLib card so much cheaper than the failed IBM effort? The answer lay in the complexity of the chips used to generate the sound. While each of the instrument channels used by the AdLib card were equipped with 2 sine-generating operators, the channels in the IBM’s card were constructed from four, greatly enhancing the fidelity and number of sounds composers were able to create. Consequently, even though the AdLib actually had an advantage when it came to simultaneous instrument sounds (AdLib had 8 vs the IBM card’s 6) there can be no denying that the individual sounds you can make on the AdLib sound a little bit thin and cheap in comparison to the sounds used in an arcade game like Outrun. If you fired up an Adlib-based tune and expected it to sound like the backing to a New Order or Depeche Mode track, you’d be in for disappointment.
Still, in the short term such drawbacks didn’t really matter: starting at $195 but quickly slashed to $175, the AdLib card was a much more affordable product when compared to it’s rivals. Though the FM chip didn’t give composers the CD-quality sound they pined for, they could at least create sounds that were evocative of brass or wood-wind instruments even if they didn’t sound quite like the real thing. Considering that the AdLib card also cemented other standards like the 3.55mm speaker ports we still use today, you think they would have become the dominant force in PC sound. Alas, that wasn’t to be.
Creative Sound Blaster/ Sound Blaster Pro
Interestingly, the story of the company who were to become the dominant force in PC sound starts with a something of an ignoble failure. Beginning life as a computer repair shop, Creative Labs quickly moved on to creating custom PCs for the Singapore market before finally expanding internationally with their Creative Music System sound card. Deploying an impressive 12 square wave generators, the card effectively sounded like a souped up version of the SN76489 used in the PCjr, albeit one that offered composers a much greater amount of harmonic versatility. Though the card was given a more exciting name for its Western debut – the Game Blaster – it was released too late. Turning up in 1988, the PC world had already been seduced by the modulated frequencies of the AdLib card, leaving no real space in the market for a beepy square-based affair.
Not to be put off, the company quickly began work on a new product they dubbed the Killer Kard, which they released in 1989. Named the Sound Blaster for its official released, it’s fair to say it lived up to its original moniker. Rather than create yet another new sound standard for DOS based computers, Creative opted to take what was there and expand it. Not only was it fully backwards compatible with Creative’s own Gameblaster, but the Sound Blaster also included the YM3812 used in AdLib’s card, making it effectively identical to the Canadian card when it came to generating fancy FM-based music. On top of this, the Sound Blaster added an all important trump card: sampled sound effects.
Looking back, the original Sound Blaster’s sampling capabilities were relatively primitive even for the time. Based around a simple Intel micro controller, the original Sound Blaster could only playback monoaural 8-bit recordings that had been sampled at half the rate of CD quality sound, and had to halve this quality again if the user wanted to record something via the microphone input. Such drawbacks didn’t really matter mind you – the AdLib had no innate ability to play samples (let alone record them) and the PC speaker wasn’t a realistic option in most cases so the Sound Blaster really was in a class of its own. Sure, the AdLib card could give you a decent rendition of the Imperial March from Star Wars, but why settle for that when the Sound Blaster could give you that AND authentic Star Wars blaster fire, TIE engine sounds and heavy Darth Vader Breathing?
While the AdLib’s low price gave it a cost advantage over other pricier sound expansions, the difference between the AdLib and the Soundblaster in 1990 was just $65 – effectively making the choice a no brainer. Worse still for AdLib, Creative rapidly followed the success of the Sound Blaster with a deluxe “Pro” model that had an extra FM chip in order to be able to play music in stereo, sampled audio at a higher frequency rate for recordings and even had CD drive support out of the box. The AdLib card may have created the sound that would define PC game music, but it was the Sound Blaster that rose to unquestionably become as the dominant force in PC audio. Though both the the original card has long since been outmoded, the line itself survived to this day, with Creative continuing to sell a number of Soundblaster-branded sound cards and headphones.
AdLib Gold 1000
(Apparently no recording exists of the Adlib Gold version of Dune 2!)
Perhaps unsurprisingly, AdLib didn’t take the rise of the Sound Blaster laying down. Just as Creative had targeted the weaknesses of the original AdLib, for their own follow-up AdLib created a card that was similar in concept to Creative’s new Sound Blaster Pro card, but made sure to beat it in all of the key areas: When it came to music, the AdLib gold packed a refined FM chip from Yamaha that was backwards with their previous designs, but which offered the composer either a larger number of sounds or the ability to create more complex (4-operator) instruments. On the sample front, the AdLib Gold knocked the 8-bit Sound Blasters off their perch by featuring a 12-bit DAC that could make sense of CD quality (16-bit, 44khz) sound. Along side this, the card also included a MIDI-compatible gameport and an optional surround-sound based add-on. The AdLib Gold was the complete package.
Unfortunately luck wasn’t on AdLib’s side – though Creative had been busy refining the original Sound Blaster and its “Pro” follow up, the company had also been working on another card, the Sound Blaster 16. This was based on the same FM chip as the AdLib Gold and also had a DAC that could handle 16-bit audio. Thanks to some set backs in testing components for the AdLib Gold (set backs it has been suggested may have been deliberate sabotage), the SB-16 was the first to market and had established a dominant lead by the time the Gold card finally launched. Within months. AdLib had filed for bankruptcy
There is definitely room for argument about which of these cards was best from a musical perspective, and they go to show just how different two sound cards could be even when built around the same brain. From a business perspective, the Sound Blaster 16 was undoubtedly the winner as its compact design made it cheaper and easier to manufacture. However this came at the cost of sound fidelity: Creative opted not to utilise an oversampling ‘buddy’ chip that Yamaha designed specifically designed to accompany and refine the sound of the OPL3 FM chip, and this made the filtering in the Sound Blaster a lot less forgiving than the AdLib’s. This means that the AdLib Gold actually responds both better and to a larger number of of sounds than the Creative card, with entire aspects of sound that would be filtered on the Sound Blaster being clearly audible on the Gold.
Either way, with AdLib’s bankruptcy the Sound Blaster brand was free to continue its march towards dominance of the computer audio market. Today, AdLib Gold cards are so rare that they can sell on auction sites for over $1000, and to my knowledge no recording of the AdLib Gold version of the Dune 2 soundtrack exists – emulated or otherwise.
The AdLib and Sound Blaster may have been quick to dominate the world of PC game music, but it’s worth noting that internal sound cards weren’t the only gig in town. For those with enough cash there was another (significantly more expensive) option: external MIDI units.
MIDI stands for Musical Instrument Digital Interface, and – rather fittingly – it originated from around the same time IBM PCs were cementing themselves in the business world. Dreamt up by a few of the big wigs in the digital instrument industry (such as Roland, Sequental Circuits, Yamaha and Korg,) it represented a surprisingly forward-thinking idea: as it was inevitable that all electronic instruments would eventually cross over into the digital realm, didn’t it make sense for them to have a standard way of talking to one another?
Formally announced by Robert Moog in the October 1982 issue of Keyboard, the arrival of MIDI changed everything. from enabling the first meaningful bedroom producers through to allowing professional musicians to effortlessly control a entire bank of keyboards on stage, MIDI marked a transformative step in the sphere of digital music. From the perspective of software developers, it had two major drawbacks, however:
- There was no way of plugging a MIDI instrument directly into a computer.
- Though MIDI worked as far as standardising the way messages were sent, MIDI didn’t cover the precise interpretation of those messages on each machine. This meant that while instrument sound 000 might be a grand piano on one machine, on another instrument 000 could be an accordion.
The first of these problems turned out to be an easy fix. Roland were quick to develop an intermediary unit – the MPU-401 – which worked in conjunction with a series of computer-specific expansion cards to allow MIDI connectivity to be added from anything from the humble Commodore 64 right up to the exotic Sharp x68000.
Consequently, had the second problem been solved quickly, the entire history of DOS-based sound may have been very different. With standardised instrument lists, PC users could have bought any midi-compatible keyboard at any price point and been assured that it would play any MIDI soundtrack correctly. However, With no such lists existing, the PC MIDI market lay open to domination by one successful trend-setter.
Having already created the MPU-401, it probably isn’t too surprising that this player turned out to be Roland. The key to their success was the MT-32, an external sound module that computer users could simply plug into their MPU-401. Though it originally retailed for $695, The MT-32 was effectively a synthesizer without the keys. This meant that it effectively held two different places in two very different markets. In the gaming sphere it was an expensive novelty, and had it been active in the gaming sphere alone it would probably have been destined to sell only a handful of copies to only the richest of gamers.However, in the musical world it was something of a bargain. $695 might have bought you 4 AdLib sound cards, but it would only have bought you half of a Roland D-50 – the flagship synthesizer that used similar technology to the MT-32. The MT-32 may never have come close to challenging internal sound cards when it came to units sold, but its relative cheapness as a synthesiser meant that the MT-32 became the defacto standard when it came to computer music, and ended up in many households that wouldn’t have spent hundreds of dollars on a humble gaming accessory as a result.
Was the MT line worth the money? In many respects yes. Based on Linear Arithmatic synthesis, the MT-32 combined partial, 16-bit samples of actual instruments with subtractive synthesis techniques that could fill-in for the unsampled parts of the tone. This was quite a clever combination, as the use of real samples allowed the sound patches on the MT-32 to sound much more realistic than those produced by via FM, while the synthesiser functions meant that composers were free to manipulate the stock samples in order to create custom sounds. It may perhaps have been slightly weaker than the FM-based competition when it came to delivering purely weird and synthetic tones, but at the time it was absolutely unparalleled when it came to emulating acoustic instruments. In 1987 the MT-32 was the closest a PC game composer could get to creating a CD quality soundtrack.
With Sierra (again – it must be said they had a pivotal role in supporting the dissemination of improved PC sound options) marketing the unit directly to gamers, Roland were quick to realise the multifaceted appeal of the MT-32. They soon released a cheaper gamer friendly version – removing unneeded features like the LCD screen and adding some generic digital sound effects – before eventually combining the MPU and MT unit into a single expansion board that simply plugged into the user’s PC. However, with even the expansion card version weighing in at over $400, the MT was destined to remain a (justifiably) costly side-show and never really entered the mainstream.
With the MT-32 family firmly embedded at the top of the PC audio market, Roland’s impetus to create a successor was spurred by updates to the MIDI spec. Almost a decade after Robert Moog’s article in Keyboard, the industry came together once again in 1991 and managed to agree on General MIDI: a new standard that formally codified things like the number of simultaneous notes a keyboard should support (polyphony), what instrument sounds a keyboard should include and what order they should be stored in. If you’ve ever wondered why every cheap MIDI keyboard can generate seashore, gunshot and bird song sound effects, General Midi is the answer. This was potentially a big deal: while soundtracks written for the MT-32 would only work on the MT-32, soundtracks utilising General MIDI would work on any compatible keyboard or synthesizer. Hurrah!
Roland had a couple of tricks up their sleeve, however. Not only did they own the samples used in the MT-32 ROM files, but they had also developed their own separate standard – dubbed GS – which sat on top of GM and added some cool additional features such as standardised digital effects and extra instrument sounds. (As a side note, pretty much everyone else who had agreed to GM had done this too. Korg’s standard was called XS, for example.) Merging all of these together, Roland’s MT follow up, the SC-55, ended up being a bit of a beast: fully General Midi-compatible, the SC-55 managed to be backwards compatible with the MT-32’s archaic MIDI standard while also packing the extra sounds and features it needed to be fully GS compliant.
How did this translate in terms of the general quality of the sound? The SC-55 is effectively an MT-32 without the smoke and mirrors. Rapidly decreasing computer memory prices meant that the ROM size could be multiple times that of the MT-32, giving Roland the space they needed to include using full 16-bit PCM recordings of many of the instruments sampled. Backed up by the range of digital effects, the sounds generated by the SC-55 were effectively CD quality. Quite literally in some cases – many of the soundtracks produced for Sega and NEC’s CD add-ons would have been recorded straight from an SC-55.
However, as the SC-55 effectively covered three different standards, mileage varied considerably. While the tracks specifically written for the features of the Sound Canvas sounded fantastic, multi-format General MIDI tracks that couldn’t use the GS sounds and features could sometimes sound a little flat. At the bottom of the pile was the SC-55’s emulation of the earlier MT-32. Unfortunately, one major feature included in the MT-32 but cut from the SC-55 was the ability to create custom sounds. Though perhaps less of an issue for a musician who was suddenly being granted access to 100s of realistic instrument sounds, it did mean that games that were written to make use of the feature inevitably sounded terrible.
Overall then, the SC-55 was like an improved MT-32 across the board. Though predictably the sound quality at the top end of the unit’s output improved on the best of the MT-32, at around $500 the Sound Canvas was still out of the target price range for your average PC gamer and, like the MT-32, it still lacked the custom sound effects found on the Sound Blaster. Though the Sound Canvas line would have a long life span in the musical sphere, like the MT-32 it would never really challenge the Sound Blaster line as a sound source for PC games.
Conclusion: ARRRGGGHH – the variety!
From the first IBM PC up to the arrival of Doom, DOS-based sound had a strange evolution. In an age where competitors Apple and Commodore had made digital sample playback a priority, the fact that the period between 1982 and 1987 saw such a paucity of innovation in the IBM-compatible arena seems bizarre. Almost as bizarre – in fact – as the huge amount of innovation that emerged in the period that followed.
Even by 1992 the story wasn’t quite over. That year saw the entry of a new competitor – Canadians Gravis with their programmable, sample-based “Ultrasound” – who were to confuse matters further in the short term. By 1995 things had largely settled down though: Alongside the burgeoning standards the Pentium Processor and Windows ’95 operating system, an advisory letter written in 1995 recommends buying nothing but Sound Blaster cards for anyone buying a new PC.
Considering that, for much of its life, there was as an equally large array of graphics and CPU options to choose from, it’s perhaps surprising that anyone opted for the unstable world of PC gaming, where the highness of outlay was matched only by the chance of ending up with a dud. Nonetheless opt for it they did. I for one are glad about that. In a world full of fixed hardware platforms that may only have had a single sound option, i find it pretty invigorating that games like Dune 2 had so many unique and interesting versions of their soundtracks. Even if that did mean they weren’t always the easiest things to install.