PS1 sample rate and audio formats, explained

Every PS1 sample rate in one table

A PS1 game could play audio four ways, and each has its own rate. The figures come from the psx-spx hardware reference.

So PS1 audio was CD quality at the output, but the sequenced music that gives the console its sound was built from samples at much lower rates.

Why instrument samples were low rate

Every instrument in a scene shared 512 KB of sound memory with the reverb, and the first 4 KB was reserved. Here's roughly how much sound fits in the rest, with the reverb off:

That's for every instrument at once, not each one. A Hall reverb takes another 43 KB. So composers recorded instruments at lower rates, kept them short, and looped a small section for sustained notes. Bright sounds such as cymbals were often kept at higher rates, and dull ones such as bass much lower.

Because the chip changes pitch by playing a sample faster or slower, any sample rate works. A 22.05 kHz sample is simply played at half speed for its original pitch. Played above 44.1 kHz it starts skipping samples, so very high notes get rougher.

How PS1 ADPCM works

The SPU can only play samples compressed in its own ADPCM format. Each 16-byte block holds:

• One byte of header, choosing one of five prediction filters and a volume shift for the block.
• One byte of flags, marking loop start and loop end.
• Fourteen bytes of data: 28 samples as 4-bit numbers. Each one says how far the next sample is from what the filter predicted from the previous two.

Twenty-eight 16-bit samples would take 56 bytes, so that's about 3.5 to 1. The cost is a faint noise that follows the signal: hard to hear on loud, busy sounds, more audible on quiet pads and long decays. That grain, plus the chip's Gaussian interpolation softening the top end, is a large part of the PS1's texture. XA streams use the same compression with the data arranged differently.

PS1 audio file formats

Sony's development tools stored audio in a handful of formats you'll see in game files and rips:

To listen to PSF rips, use a PSF player such as the foo_psf plugin for foobar2000. To turn a game's SEQ and VAB files into MIDI and a SoundFont, use VGMTrans, as covered in PS1 SoundFonts. jPSXdec extracts XA and STR audio as WAV files. Ripped audio belongs to the game's publisher, so it's fine for study and practice, not for music you release.

Getting the sound in your DAW

You don't need real PS1 files to get the sound. Copy the sample rates instead:

1. Resample your instruments to 22.05 kHz, or 11.025 kHz for basses and soft pads, before they go into your sampler.
2. Add a little grain with a gentle bitcrusher (10 to 12 bits), rather than crushing the whole mix.
3. Soften the top with a low-pass filter around 8 to 12 kHz on each instrument.
4. Leave the master at 44.1 kHz. The PS1's output was clean. Only the samples and the reverb were low rate.

Or use simpleSPU, which stores its samples as real 4-bit ADPCM and plays them with the chip's Gaussian interpolation and half-rate reverb. Its Crush knob pushes the ADPCM grain further. The full method is in how to make PS1-style music.

Common questions

Was PS1 audio CD quality?

The output was: 16-bit stereo at 44.1 kHz. CD audio tracks were full quality too. Sequenced music was built from compressed samples at lower rates, so it wasn't, even though it came out of a CD-quality mixer.

What bit depth did the PS1 use?

16-bit for the output, the reverb and CD audio. Samples were stored as 4-bit ADPCM and decoded to 16-bit as they played.

What sample rate is XA audio?

37.8 kHz or 18.9 kHz, in mono or stereo. The lower rate let a disc fit more streams, such as dialogue, side by side.

Did the N64 have a higher sample rate than the PS1?

Usually lower. N64 games chose their own output rate, and many ran well below 44.1 kHz. See PS1 vs N64 audio.

More PS1 guides

• The PS1 sound chip
• PS1 vs N64 audio
• PS1 reverb
• PS1 SoundFonts
• How to make PS1-style music
• All guides