From soundchips to samples: How the demoscene learned to play real sounds

For the first decade of home computer music, "composing" meant programming a chip. On machines like the Commodore 64 or the early Atari 8-bits, a musician didn't record anything — they described a sound mathematically, as a waveform, a frequency, and an envelope, and the chip synthesized it live, every time the program ran. That approach produced brilliant, characterful music, but it was fundamentally limited: a square wave is a square wave, a triangle is a triangle, and getting a chip to sound like a real drum, a human voice, or an orchestral instrument required enormous cleverness and, even then, only ever approximated the real thing. The single event that broke this limitation open, and set the demoscene on an entirely new path, was the arrival of sample-based music — and the piece of software most responsible for popularizing it was Karsten Obarski's Ultimate Soundtracker.

1. Before samples: synthesis only

Chips like the Commodore 6581 SID or the Atari/AY-family YM2149 generated every note from scratch using oscillators, envelopes and (sometimes) noise generators — see our earlier articles on those chips for the technical detail. Composers on these systems were, in a very real sense, sound designers first and melody writers second: getting a convincing "bass drum" out of a SID's noise channel and a fast envelope was a genuine skill in itself, distinct from writing the tune. This was expressive, but it was a closed world — the chip's built-in waveforms were the entire sonic palette available, and no amount of clever programming could make a square wave sound like a sampled guitar pluck or a spoken word.

2. The game-changer: PCM playback and the Amiga

The breakthrough came from a completely different design philosophy, embodied by the Commodore Amiga (1985) and its custom audio chip, nicknamed Paula. Rather than synthesizing sound from oscillators, Paula was built to play back digitized audio samples directly: four independent channels of 8-bit PCM (pulse-code modulation) audio, mixed and resampled entirely in hardware via DMA, completely independent of the CPU. In practice this meant a composer could record — or "sample" — a real sound (a plucked bass string, a drum hit, a vocal snippet, anything at all) and simply play that recording back at different pitches to form a melody, instead of describing the sound mathematically to a synthesis chip.

As one detailed comparison of Atari and Amiga sound architecture puts it, the two philosophies were almost perfect opposites: Paula worked like "a basic PCM rompler," while the Atari's YM2149 worked like "a basic PSG synthesiser" — two very different tools built to do a similar job.

3. Karsten Obarski and the arrival of Ultimate Soundtracker (1987)

The Amiga's hardware made sample playback possible, but it took a dedicated piece of editing software to make composing with samples practical for ordinary users rather than professional programmers. That software was Ultimate Soundtracker, written by German developer Karsten Obarski for the games company EAS and released commercially in December 1987. Obarski had previously worked with Chris Hülsbeck's C64 program SoundMonitor, and borrowed its vertical, scrolling, column-based layout — an interface idea that would go on to define the entire tracker genre.

Ultimate Soundtracker let a musician load short digitized sound samples into memory, arrange them across Paula's four channels at up to roughly 29 kHz playback, and sequence them note-by-note in a grid-based "pattern," where each row represented a moment in time and each column represented a channel. It saved its songs in what became known as the MOD format — a file that bundled both the small raw samples and the note/pattern data together, making entire songs remarkably compact. Crucially, this was the first music software to use this sampled track-sequencing approach, and it is literally where the word "tracker" comes from.

4. Why this mattered so much to the demoscene

Before Ultimate Soundtracker, writing computer music required real programming skill: composers on the C64 or Atari 8-bit worked directly with assembly-level sound routines or specialized tools built by programmers for programmers. The tracker model changed the economics of music-making completely. A tracker was, in effect, a self-contained sequencer and sample-arranger: no assembly code, no synthesis theory, just samples, notes, and a grid. This lowered the barrier to entry so dramatically that entirely new communities of musicians — many with no formal programming or music-production background at all — began producing music for demos within a few years of the Amiga's release.

It also permanently changed what "computer music" could sound like. Suddenly a demo could feature a sampled bassline that sounded like an actual bass guitar, a drum break sampled from a real record, or a vocal snippet — sounds that had been completely out of reach for pure-synthesis chips. The demoscene, previously defined almost entirely by chiptune-style square-wave melodies, began to sound noticeably closer to the sample-based dance and electronic music being produced in professional studios of the same era.

5. A tale of two 16-bit machines: Amiga vs. Atari ST sampling quality

The Amiga's biggest 16-bit rival, the Atari ST, took the opposite approach at launch: it shipped with the YM2149 synthesis chip and no dedicated sample playback hardware at all. Sample-based tracker music was still possible on an original ST, but only through pure software trickery — the CPU itself had to twiddle the YM2149's volume register extremely rapidly to approximate a digital waveform, a technique that ate up precious processing time and produced noticeably rougher, noisier results than the Amiga's dedicated hardware mixer.

Atari eventually responded with the STE revision, which added genuine 8-bit stereo DMA sound hardware, supporting playback at four fixed rates up to roughly 50 kHz. On paper this looked competitive with, or even superior to, the Amiga's specs. In practice, however, forum discussions among Atari and Amiga users highlight a persistent gap: because the STE could only output at a handful of fixed sample rates rather than Paula's per-channel variable rate, playing back a normal MOD file (composed with arbitrary pitches) meant the CPU had to resample and remix every note into one of those fixed rates in real time, burning far more CPU time than the Amiga's dedicated hardware mixer needed. One widely shared technical comparison summarizes it plainly: on the Amiga, "audio mixing and resampling are done in hardware… the CPU is only responsible for selecting a waveform and its length, the pitch, and the volume," whereas the STE needed real software mixing for the same result.

The upshot, echoed repeatedly by musicians who used both machines at the time, is that the STE's DMA sound chip, while a big improvement over the plain ST, "is still not as good as the one in [the] Amiga." Skilled ST/STE musicians and later trackers such as MaxYMiser and Octalyser did coax impressive multi-channel digital sound out of the hardware, sometimes blending YM chiptune channels with DMA sample channels for a hybrid sound — but the underlying architecture never quite matched the Amiga's effortless, CPU-free sample fidelity.

6. The PC's slow, bumpy road to decent samples

While Amiga (and to a lesser extent, Atari ST) musicians were already trading sample-based MOD files in the late 1980s, the IBM PC was years behind. Early PC "sound cards" like the AdLib (1987) offered only FM synthesis with no sample playback at all. The original Sound Blaster (1989) added an 8-bit mono PCM channel, but at low, coarse sample rates — Creative Labs' own co-founder later admitted plainly that "the audio quality was coarse and very bad by today's standards", even though it was a huge leap from having no digital sound at all.

Through the early 1990s, PC sample quality improved in fits and starts:

The Gravis Ultrasound deserves special mention here (and is covered in depth in our article on FastTracker): it was arguably the first PC sound card that could genuinely match the Amiga's trick of mixing many sample channels entirely in hardware, which is exactly why it became the preferred card of the PC demoscene almost overnight. As one retrospective on PC audio history put it, games and demos using the GUS could play "quality Amiga-style sample based audio on a 386, when you'd need a Pentium to get the same performance using software-based mixing with a regular Sound Blaster card."

It wasn't until the Sound Blaster 16 (1992) that the PC could even match 16-bit, 44.1 kHz "CD quality" sampling on paper — five full years after the Amiga's Ultimate Soundtracker had already popularized sample-based composing at 8-bit resolution. And it took the arrival of onboard-RAM wavetable cards like the AWE32, plus the general explosion in CPU power through the mid-to-late 1990s, before PC musicians could routinely work with sample libraries as rich and varied as what Amiga composers had enjoyed for years already.

7. From MOD to XM to beyond

As sample-based tracking matured, the amount of detail composers could control kept growing: from the Amiga's original 4-channel, 8-bit-only MOD format, to PC-era formats like FastTracker 2's 32-channel XM format with volume/panning envelopes, and eventually to fully 16-bit, CD-quality sample formats used by later trackers and DAWs. But the underlying idea — arranging short recorded sounds on a timeline instead of describing a waveform to a synthesizer — traces directly back to Obarski's original Ultimate Soundtracker, and to the Amiga hardware that made it possible in the first place.

It's a nice historical footnote that the demoscene, having pioneered sample-based composing on inexpensive home computers years before professional studios' digital samplers became affordable, effectively invented a cheap, DIY version of a technique the music industry itself had only recently begun exploring with costly hardware samplers like the Fairlight CMI or the E-mu Emulator.

Further Reading

• Wikipedia: Ultimate Soundtracker
• "Soundtracker origins, part 1: Where in the World is Karsten Obarski?" — detailed history
• Introduction to Demoscene: Chip music as hacker culture
• Wikipedia: MOD file format
• Wikipedia: Commodore Amiga
• Atari ST vs Amiga: sound architecture compared
• Atari-Forum: "Audio vs Amiga" discussion thread
• "The History of PC Audio"
• "The Sound Blaster Story" — Custom PC
• Wikipedia: Chiptune

Read more stories →