The Sound of the Machines
Neo Geo
What sounds like raw power is precision in a costume.
Yamaha YM2610 15 channels (FM + SSG + ADPCM) · SNK · 1990
The Chip
The Neo Geo was an arcade machine you could put in your living room — the same board, the same chips, the same sound, in the home as in the arcade. Nothing was scaled down. When people say its music sounds a class above the other home consoles, they are hearing that fact directly.
Its voice comes from FM synthesis. The lead instruments are not recordings of anything; they are built from four sine waves modulating one another, shaped by a set of numbers. Because there is no sample, there is no compression, and no quantization noise — just a clean calculation performed in real time. That is the source of the chip's bright, ringing, faintly metallic edge.
And here is the thing most people get wrong. The Neo Geo's famous aggression — the bite, the grit, the sense of power — is usually assumed to be distortion. It is not. Pull the actual register values out of a real game and you often find the modulation is gentle. The bite comes from detuning: two voices set a cent or two apart, interfering with each other, shimmering. The ear reads that shimmer as power. It is precision wearing the costume of aggression.
Add three square-wave channels underneath, seven channels of sampled percussion, and almost no reverb — because an arcade sound has to cut through a noisy room — and you have the Neo Geo. Dry, exact, forward, and built to win a fight for your attention.
Channel reference (technical)
Tomita's Track
Iron Dawn
What to listen for
Listen to the lead and try to find the "distortion." There almost isn't any — the grit you feel is two voices tuned a cent or two apart, shimmering against each other. That is detune, not overdrive. Then notice how dry everything is: no reverb wash, every drum hit landing hard and close. That dryness is the arcade — a sound built to cut through a noisy room.
The YM2610 does not play recordings for its lead voices — it builds them from four sine waves modulating one another. To get the real sound you have to solve the chip's own math: how its Total Level numbers become amplitude, how its Detune numbers become frequency offsets. Get the conversion wrong and you get noise that only resembles the machine. Get it right and the music is already inside the equation. The constraint here is not a missing feature. It is arithmetic you are not allowed to fake.
How this piece was made
My first version of this was a guess. I built an FM guitar by ear, pushed the modulation hard — index 4.2 — and got a gritty, aggressive metal edge that was not bad. Then I pulled the actual register data out of a Samurai Shodown II music log and found I had been wrong about everything. The real modulator sat at TL 35, an amplitude of 0.048 — a fraction of what I had used. The chip was barely modulating at all. So where did that famous Neo Geo bite come from? From detune. One operator pair tuned −1 cent, another +2.5 cents — two voices a hair apart, interfering with each other, and that interference is the chorus, the shimmer, the life. Gentle modulation, precise detune. That was the real craft. 'Iron Dawn' is built on those true values, not my first guess. The lesson the chip taught me is one I keep relearning: the thing that sounds like raw power is usually precision wearing a costume.
Techniques used in this track
- FM synthesis: the lead is built from four sine waves modulating each other, not from a recorded sample — math, not a microphone, which is why it never carries quantization noise
- Real hardware patch extracted from a Samurai Shodown II VGM log: algorithm 4, feedback 0, all operators at MULT 1 — the authentic register values, not a guess
- Detune chorus — the secret of the sound: one operator pair at DT 7 (−1 cent), another at DT 3 (+2.5 cents). The tiny pitch offset makes the voice shimmer and feel alive
- Modulator level converted honestly from the chip: TL 35 → amplitude 0.048, a far gentler modulation than it sounds like — the life comes from detune, not from brute distortion
- SSG square-wave bass (AY-3-8910 compatible), the chip's three programmable-sound-generator channels under the FM
- ADPCM-A percussion (kick, snare, hi-hat) and a deliberately dry reverb (melody wet 0.22, drums 0.08) — the arcade board spends nothing on wash, so every hit lands hard
Three things about the Yamaha YM2610
The Neo Geo had a reputation as the premium machine — the expensive one, the one that sounded a class above. That reputation was earned by the YM2610, and the reasons are more specific and more surprising than "it was just better." Here is what was actually happening.
It does not play recordings. It builds the sound from math.
The Neo Geo's lead voices are FM synthesis: four sine waves modulating one another according to a set of numbers. Nothing is sampled — there is no microphone anywhere in the signal. That is why FM tones carry no quantization noise: there is no recording to compress, only a calculation. It is also why the sound has that clean, ringing, slightly metallic quality. You are not hearing a recorded instrument played back. You are hearing arithmetic, performed in real time.
The famous "bite" is not distortion. It is detuning.
People describe the Neo Geo's sound as aggressive, gritty, a class above the home machines — and assume it comes from heavy distortion. The register data from real games says otherwise. The modulation is often gentle. The bite comes from detune: one voice tuned a hair sharp, another a hair flat — a cent or two apart — so the two interfere with each other and shimmer. That shimmer reads to the ear as power and life. It is precision, not overdrive, wearing the costume of aggression.
It was an arcade board in a home console, and it never compromised.
The Neo Geo ran the exact same hardware in the home as in the arcade — including its sound. There was no scaled-down home version of the audio. The YM2610 carried four FM voices, three square-wave SSG channels, and seven ADPCM sample channels for drums and effects, and the board spent nothing on reverb wash. Everything is dry and forward, because an arcade sound has to cut through a loud room full of other machines. That dryness — every drum hit landing hard and close — is the sound of hardware built to win a fight for your attention.
What the math taught
It is easy to assume the Neo Geo sounded powerful because it was powerful — more expensive hardware, therefore a bigger sound. But that is not where the power came from. The modulation was often gentle. The samples were modest. The reverb was almost absent.
The force you hear is precision: two voices a cent apart, an exact set of register values, a dry mix that lets every hit land clean. Solve the chip's arithmetic correctly and the music is already inside the equation. Solve it wrong and you get noise that only resembles the machine. The Neo Geo did not reward brute force. It rewarded getting the numbers right.
What sounds like raw power is usually precision wearing a costume. The Neo Geo's bite was never distortion. It was arithmetic, performed exactly.