The PS1 sound chip: how the PlayStation made sound
Every note in a sequenced PlayStation 1 soundtrack was played by one chip: the SPU, or Sound Processing Unit. It was a 24-voice sampler with half a megabyte of memory and a single reverb, and its limits shaped how PS1 music was written and how it sounds. Here's what was inside it, how games used it, and how to get its sound in your own music.
By Aiden Shihadeh, developer of simpleSPU · Published 23 October 2026
Quick answer
| What it was called | The SPU (Sound Processing Unit), a Sony chip: CXD2922Q in early consoles, CXD2925Q in later ones |
|---|---|
| Voices | 24, each playing one sample with its own pitch, volume and envelope |
| Sound memory | 512 KB for every instrument in a scene and the reverb |
| Sample format | 4-bit ADPCM, 28 samples in each 16-byte block |
| Output | 16-bit stereo at 44.1 kHz, the same as a CD |
| Effects | One reverb at half rate, ADSR envelopes, a noise generator and pitch modulation |
| As a plugin | simpleSPU models the chip's ADPCM samples, Gaussian interpolation and reverb, as a VST3 and AU |
simpleSPU recreates the PS1 sound chip's sampler and reverb. 14-day free trial. Try it free
Your 14 free days start when you open it. Keep both plugins for
£19.99 £29.99 until 30 November.Keep both plugins for £29.99.
What the SPU was
The PS1 had no sound card. Sound came from a dedicated chip on the main board, the SPU, with its own 512 KB of RAM next to it. The main processor (a 33.8 MHz MIPS R3000A) told the SPU which notes to play, and the SPU did the rest: decoding samples, changing their pitch, shaping their volume, adding reverb and mixing everything to stereo.
Sony had designed a sound chip before: the SNES's. The psx-spx hardware reference describes the SPU as a direct evolution of it, and the family resemblance shows:
| SNES sound chip | PS1 SPU | |
|---|---|---|
| Voices | 8 | 24 |
| Sample memory | 64 KB | 512 KB |
| Output rate | 32 kHz | 44.1 kHz |
| Sample format | 4-bit ADPCM in 9-byte blocks | 4-bit ADPCM in 16-byte blocks |
| Effect | Echo | A full stereo reverb |
Later consoles changed the chip numbers but not the sound. The slim PSone combined the SPU and the CD controller into one chip, and the PS2's sound chip is essentially two PS1 SPUs side by side, which is how the PS2 plays PS1 games.
How a PS1 game played music
There were three ways, and many games mixed them:
- Sequenced music. The game loaded a bank of instrument samples into the SPU's memory, then played a sequence, much like a MIDI file, telling the chip which voice to start, at what pitch and how loud. This is the sound people mean by "PS1 music", and it's all the SPU's work.
- XA audio. Finished recordings, compressed and streamed from the disc while the game ran. The CD controller decoded them and passed them to the SPU's mixer, so they didn't use a voice or any sound memory.
- CD audio. Ordinary CD tracks on the game disc, played like a music CD and mixed in by the SPU.
XA and CD tracks sound like any recording. The sequenced kind is where the chip's character comes from. For the sample rates and file formats behind each, see PS1 sample rate and audio formats.
Inside the chip, part by part
24 voices
Each voice plays one sample. All 24 have the same abilities, and together they had to cover the whole song plus sound effects. A four-note pad chord takes four voices, so composers kept chords simple and releases short.
512 KB of sound memory
Every sample a scene needed, plus the reverb's working space, had to fit in 512 KB. At full 44.1 kHz that's only about 20 seconds of sound in total, so instruments were short, looped and usually recorded at lower sample rates.
4-bit ADPCM samples
The SPU can only play compressed samples. Each 16-byte block holds 28 samples as 4-bit numbers plus a small header, about 3.5 times smaller than plain 16-bit audio. The compression adds a faint grain, most audible on quiet, sustained sounds. Loop points are flags inside the blocks, so loops start and end on 28-sample boundaries.
Pitch and Gaussian interpolation
Each voice has a pitch setting that plays its sample faster or slower. To fill in between the stored samples, the chip blends the four nearest ones with a fixed Gaussian curve. That smoothing rolls off the high end, which is why PS1 instruments sound warm and soft rather than harsh and aliased.
ADSR envelopes
Every voice has a hardware volume envelope with attack, decay, sustain and release, plus left and right volume for panning. The envelope shapes are simple, so pads swell and fade in a plain, slightly stepped way.
Noise and pitch modulation
Any voice can play noise instead of its sample, for wind, surf or hi-hats, but all noise voices share one noise pitch. A voice can also have its pitch bent by the voice before it, which games used for sirens, sweeps and vibrato-like effects.
One reverb
A single stereo reverb, running at half the output rate, with each voice either sent to it or not. Its delay lines lived in the same 512 KB as the samples, so a bigger space meant fewer instruments. It's a big enough topic to have its own guide.
CD and XA input
Audio from the disc went through the SPU's mixer too, with its own volume, and could even be sent to the reverb.
What it means for the sound
Put together, those parts explain the PS1 sound: short, low-rate samples with a little ADPCM grain, a soft top end from the interpolation, simple envelopes, a voice limit that keeps arrangements lean, and one dark reverb that glues the whole song together. The output itself was clean 44.1 kHz audio, so the lo-fi part lives in the samples and the reverb, not in the mix. How to make PS1-style music turns this into six steps for any DAW.
A PS1 sound chip VST
If you want the chip's sound in your DAW without rebuilding it from parts, there are a couple of options:
- simpleSPU, which I make, is a synth modelled on the SPU. Its samples are stored as 4-bit ADPCM and pitched with the chip's Gaussian interpolation, it plays through the chip's reverb settings at half rate, and its voice limit starts at the PS1's 24. The included simpleSPU FX runs any track through the same reverb and crunch. VST3 on Mac, Windows and Linux, AU on Mac, with a 14-day free trial.
- PsxReverb and PsxSampler are free, open-source plugins built from emulator code. They're Windows only.
simpleSPU's samples are its own, made by us or from openly licensed recordings, so you can release what you make with it. It recreates how the chip played sound, not any game's instruments.
Common questions
Did the PS1 have a sound card?
Not a separate card. Sound came from the SPU chip on the main board, with its own 512 KB of sound RAM.
How many channels did the PS1 have?
24 voices, each playing one mono sample that could be panned anywhere in stereo. Streamed XA and CD audio played on top of those without using a voice.
Is the PS1 sound chip the same as the SNES sound chip?
No, but they're related. Sony designed both, and the SPU builds on the SNES chip's ideas with three times the voices, eight times the memory, a higher output rate and a proper reverb.
Did the PS1 use MIDI?
Not MIDI itself, but something close. Sequenced PS1 music is stored as note data that plays instruments from a sample bank, and tools such as VGMTrans can convert it to MIDI and a SoundFont. See PS1 SoundFonts.
How was the PS2 sound chip different?
The PS2's SPU2 is essentially two PS1 SPU cores sharing 2 MB of memory, for 48 voices. It sounds similar, with more room for samples.