Before I actually talk about PC audio itself, it’s worth mentioning a couple of important dates in audio history, pre-PC.
The first really practical application of stereophonic sound was patented by record company EMI in the 1930s, aimed at creating stereo recordings for music playback using vinyl record. The actual implementation didn’t happen until the early 1950s, when LP and 45 RPM stereo records started being sold.
Stereo became the norm in home audio systems for decades, despite the ill-fated consumer experiment with quadrophonic sound in the early 1970s. During the 1970s and 80s, Dolby Labs began experimenting with multichannel audio for film. This wasn’t for home audio, but for movie theaters. Most of the early efforts were analog, using magnetic stripes on the film itself. Later, Dolby created a stereo optical format. Despite the use of “stereo”, this was actually multichannel audio, supporting four or more channels, but was known as stereo because there were no surround or rear channels.
In the late 1980s, Dolby Digital was born, and fully digital soundtracks for both movies and, later, DVDs came to pass. People everywhere become more comfortable with multichannel audio, even the matrix versions used in techniques like Dolby ProLogic and its successors. Digital Theater Systems (DTS) became Dolby’s leading competitor, particularly in home theater audio reproduction.
Now let’s rewind to the 1960s. Researchers at NASA began experimenting with the idea that sounds arrive at your ear from different places at different times. The shape of your ear has an effect on how those different sound waves are channeled into your inner ear. Your brain interprets the delays and directionality to build positional cues, so you can figure out where and how far away a particular sound is.
They took this idea a step further, using some sophisticated math to model those sound delays, shaping the audio using something called “head related transfer functions” or HRTFs. In other words, different audio streams that may come from a single source could be manipulated to seem to the listener like they were coming from different directions and distances.
Years later, HRTF for positional audio began arriving for PCs; we’ll talk about that shortly. Okay, let's get on with the list.
When the IBM PC arrived on the scene in the early 1980s, its CGA graphics option could, in theory, handle low resolution games. But unlike its home computing competitors from Apple, Commodore, and Atari, there was no provision for any kind of audio beyond diagnostic beeps. That’s probably because IBM’s original target for its first PC was small businesses.
Game developers and gamers wanted better sound from their games. Along came Ad Lib, Inc., founded by a former music professor named Martin Prevel. Ad Lib, a Canadian company, was the first hardware manufacturer to ship a dedicated add-in PC sound card: the Ad Lib Music Synthesizer.
Ah, the Ad Lib. Unfortunately, first doesn’t always mean forever.
The Ad Lib used the Yamaha YM3812, an FM synthesizer chip. FM synthesis builds sounds in a completely procedural manner, similar to the early Moog synthesizers, rather than using digitized samples of actual audio. Good FM synthesis can actually generate reasonably good quality music. The real problem with the Ad Lib was the complete lack of any digital audio support. That precluded the use of voice samples for dialog, for example.
The first game to actually support the Ad Lib was Sierra On-Line’s King’s Quest IV. Our first encounter with the Ad Lib was in an early LucasArts flight sim, Their Finest Hour: The Battle of Britain. The game used FM synthesis for sound effects (gunfire, engine sounds), and the level of immersion was startling. So being a classic early adopter, we ordered an Ad Lib after watching the demo.
So, what happened to Ad Lib? Sadly, the company made one of the classic errors of technology startups: it didn’t respond in a timely manner to new competition. That competition came from a company familiar to gamers everywhere: Creative Labs, and the Sound Blaster card. In 1992, Ad Lib filed for bankruptcy, while Creative’s Sound Blaster family continues to exist today.
Sierra On-Line’s Ken Williams wanted better music in Sierra’s line of adventure games. So the company hired musicians to compose lush soundtracks, and then included MIDI files of those soundtracks in the games. MIDI (Musical Instrument Digital Interface) files consist of short messages that tell a particular piece of hardware the pitch, intensity, and other information about a particular sound. It’s up to the hardware to decode the file into something resembling music.
Imagine what Sierra’s games would have been like without those cheesy MIDI scores.
To better facilitate music, Sierra On-Line sold a Roland MT-32 and Roland LAPC-I MIDI synthesizer hardware. The MT-32 was a rather bulky external box, while the LAPC-I was a very long, 8-bit ISA card. Other game companies began supporting the higher quality Roland MIDI for games. I used an LAPC-I for about a year, and the music it generated was certainly a cut above what the Yamaha OPL chips built into the various Sound Blasters (and the original Ad Lib) could generate. But the LAPC-I was over $200, and the MT-32 was a cool $550–very large chunks of change just to get a better soundtrack.
The prohibitively high cost of the Roland hardware turned a lot of users off. When Creative Labs shipped the Sound Blaster 16 in 1992, it included a connector for daughter cards with built in wavetables capable of delivering higher quality MIDI. A mini-industry boomed for a short period of time, with different companies selling MIDI daughter cards for the Sound Blaster 16 and similar cards.
MIDI daughter cards attempted to bring higher quality music to PC gaming
Eventually, wavetable support was built into the basic sound card itself, and later into Windows, so the need for additional external hardware faded away.
As we noted, Creative Labs dominated the sound card business by releasing countless products, some very similar to others. That didn’t prevent a number of companies from trying to compete. Some competed on the basis of compatibility with the Sound Blaster de-facto standard, while others tried to pioneer different approaches. We can’t possibly touch on all competitors, but let’s talk about some of the more important ones. Some, like Ensoniq, offered interesting innovations, but never really had much market presence.
The Gravis Ultrasound
Perhaps the most interesting alternative to the Sound Blaster was the flawed, but innovative, Gravis Ultrasound and its successors. Developed by Canadian company Advanced Gravis, the Ultrasound’s real claim to fame was its high quality music authoring and playback capabilities. It was the first consumer level sound card to support MIDI wavetable synthesis.
The Ultrasound’s music playback capability was astonishingly robust for the time. This being the era of DOS gaming, the Ultrasound tried to position itself as an alternative to Creative. Back then, developers had to either use third party middleware to talk to sound cards, or write directly to the hardware. The Ultrasound appealed to quite a few game developers, but it proved to be a difficult platform to develop for.
Gravis also attempted to build compatibility with the Sound Blaster by emulation, through the use of a TSR (terminate-and-stay resident) known as “Mega-Em.” This particular piece of software was a pain to run, and caused problems with games that consumed large amounts of system RAM and often needed to use extended memory on their own.
Media Vision arrived on the scene in the early 1990s, replete with substantial venture capital funding. The original Pro AudioSpectrum card featured ostensibly better hardware than Creative Labs, but always had minor compatibility problems that didn’t make them 100% Sound Blaster compatible. Media Vision also helped drive the multimedia revolution, being one of the first companies to bundle a sound card with a CD-ROM drive, and even helped develop and sell several multimedia games.
Perhaps the most interesting aspect of Media Vision was its meteoric growth and equally rapid tumble, due to a massive financial scandal due to securities fraud charges against the founders. According to Wikipedia, this securities fraud case cost investors and bond holders $200,000,000. Eventually, Media Vision sold off all its assets and inventory, and morphed into a new company: Aureal Semiconductor.
While Aureal Semiconductor rose from the ashes of Media Vision, it was a very different company. Its business model was similar to graphics chip companies: it was a fabless chip company that designed its audio chips, farmed the manufacturing out to external fab facilities and sold the chips to other companies to make boards.
The first chip was the Aureal Vortex, but its real secret sauce was A3D. A3D was an early implementation of HRTF (hear related transfer functions), which could procedurally synthesize positional audio. A3D was built on the original NASA research into HRTFs. While HRTFs aren’t perfect, the tech as it pertained to PC gaming was better than anything else that existed at the time. Game developers like LucasArts and iD software engineered A3D support into their titles.
Aureal eventually fell victim to patent infringement litigation filed by Creative Labs. Despite winning all claims brought against it, the cost of the litigation drove Aureal out of business, and the company’s intellectual property and assets were, in the ultimate irony, bought by Creative Labs.
Even though the Sound Blaster de-facto standard owned a vast majority of the market, game developers were growing tired of writing to bare metal. While middleware suppliers existed, they tended to be small and lacked the ability to support more than a handful of existing audio suppliers.
Windows 95 arrived in summer of 1995, and along with Windows 95 came the promise of DirectX. DirectX was a fully supported middleware layer built into Windows, supported by Microsoft and aimed at the hearts and minds of game developers. DirectX offered an abstraction layer for underlying hardware, enabling game developers to write to a single, albeit sometimes complex, API. This API included DirectSound. Windows 95 also brought a standardized software-based wavetable based on Roland’s implementation of the 128-voice general MIDI sample set.
It’s hard to believe that DirectX and Windows 95 happened 15 years ago.
DirectX and Windows laid the groundwork for what would eventually spell the end of sound cards as a required piece of hardware for PCs and PC gamers. While sound cards are still sold, they’re no longer an essential part of the PC gaming mix. Why did sound cards become more of a luxury accessory than a necessity?
Two things happened: multicore, high performance CPUs, and bad drivers.
Real-time, multichannel positional audio processing could consume significant amounts of a CPU’s available horsepower in the era of single core processors. But as PC processor floating point capabilities became more robust, core clock frequencies increased, and dual and quad core processors became commonplace, the CPU could handle most of the audio processing chores. The need to offload audio processing to an external DSP (digital signal processor) was no longer a necessity.
Creative Labs had made some attempts to enter the sound card business with something called the Creative Music System. The CMS was later relabeled “Game Blaster”, but it made little headway in either market. The Sound Blaster arrived on the scene in 1989, using the same Yamaha FM synthesizer chip as the Ad Lib.
However, the Sound Blaster also included the capability to play back digitized sound samples. The actual audio quality was pretty poor: monaural, 23KHz sampling frequency, with recording sample rates at up to 12kHz (roughly AM radio quality). It was able to decompress ADPCM encode audio, so game developers could ship sound files that were fairly compressed. Compression was a good thing – this was back in the day when most games still shipped on floppy disks.
The AWE64 was the last ISA sound card from Creative Labs.
The Sound Blaster Live was an early effort to support multichannel PC audio.
In addition to the ability to handle digital audio, the Sound Blaster had an IBM-compatible analog game port, making it a one-stop solution for gamers. While Ad Lib was trying to build its response – the Ad Lib Gold – the Sound Blaster pretty much swept the market, garnering not only support from gamers but from game developers, as it allowed them to (finally) add digital audio effects to their titles.. Since the Sound Blaster was completely compatible with the Ad Lib, gamers could buy it and feel confident their existing games would work.
One of the first games to support the Sound Blaster’s digital audio capabilities was Wing Commander II, in the form of an expansion pack consisting of seven floppy disks. Those floppies contained all the digitized audio for in-game voice. So you not only got music, but your wingmen and opponents would talk to you, and some of the cut scenes were voiced as well.
The Sound Blaster went through numerous product iterations over the years, including multiple similar versions of some products. Eventually, Creative added support for digital signal processors to offload audio processing from the PC, integrated MIDI wavetable support, and support for newer PC interfaces (16-bit ISA, then PCI and, eventually, PCI Express). The steady stream of product releases, plus aggressive marketing and occasional litigation against competitors kept Creative Labs on top of the sound card market for two decades.
As Windows XP arrived in 2001, Microsoft built into XP the capability to monitor events -- particularly system and software crashes -- and upload them to a centralized database with the permission of the user. The boys in Redmond collected vast amounts of data on system problems over XP’s lifetime. They also collected data from its support centers, as users called in to get help with various problems.
It seemed that audio hardware -- particularly sound cards with onboard hardware acceleration -- accounted for more problem issues than any other single piece of hardware. Even graphics drivers, often the bane of PC gamers, didn’t create as many problems.
So when Microsoft developed Windows Vista, it designed Vista with native support only for software audio solutions. Every motherboard has basic codec (compressor-decompressor) hardware built into it; this was required by the AC97 (Audio Codec ’97) standard developed for PCs jointly by Microsoft and major hardware suppliers. The company reasoned that given all those excess CPU cycles, it could remove one complication by removing built-in support for hardware accelerated audio.
The sound card industry – now mostly driven by Creative Labs – responded by developing OpenAL, an open audio API aimed mostly at game development. OpenAL is useful because of its operating system independence, and quite a few modern games support it. But OpenAL also has a fallback for systems without separate audio hardware, meaning that hardware acceleration isn’t needed.
Did the Blue Screen of Death ultimately led to the downfall of the add-in soundcard?
It definitely played a role.
Despite Microsoft’s efforts to marginalize sound cards, Creative Labs developed perhaps the most robust PC sound card chip developed to date, the E-Mu 20K1, aka the X-Fi. The X-Fi was built on a 130nm manufacturing process, had 51 million transistors and a robust DSP architecture that could handle most types of PC audio. Creative had taken the lessons of its earlier driver problems to heart, and the X-Fi drivers were generally more robust than previous Creative drivers.
In some respects, though, the X-Fi and sound cards built on it, plus Creative’s revamped software effort, was too little, too late. On top of that, Creative didn’t jump on the PCI Express bandwagon for too long, and the PCI Express X-Fi cards ended up being pretty late to the party. It’s likely that history will look at the X-Fi as being interesting, but somewhat irrelevant, technology in the history of PC sound.
Most modern motherboards, particularly midrange and high end retail motherboards used in PC game systems, mostly support sound through CPU processing and output through simple hardware codecs. Since sound cards are no longer required, they’ve moved up the spectrum, offering richer capabilities but at generally higher prices. In the end, the PC sound card, which helped enriched gaming experiences for PC gamers since the original Ab Lib, are now really optional items for audiophiles and audio geeks rather than a necessary add-on just to hear high quality sound from your PC.