← Gazillion Industries/Reading material

The SP-303 vinyl sim: the sound, the compressor, and a free way in

Gazillion Industries · September 20, 2026 · updated October 28, 2026

A plastic box with eight rubber pads, no battery compartment and no microphone, sold in 2001 as the cheapest way into sampling. It could hold thirty-one seconds of audio at full quality. Two of the most influential records of the following decade came out of it.

The BOSS SP-303 Dr. Sample is a compact phrase sampler released in 2001, best known for one effect: the Vinyl Simulator, which adds the noise, the band-limiting and the squash of a record to whatever you put through it. Producers used it as a sound rather than a studio, and its reputation now rests almost entirely on what it does to audio on the way in. Searched as the 303 vinyl sim, the part most people are after is the sp303 vinyl sim compressor: the squash under the noise.

The short version

What the machine actually was

It replaced the SP-202, and by the spec sheet it was a downgrade in two obvious ways: the built-in microphone and the battery power both went. What it gained was the part that mattered. Sound On Sound's 2001 review noted the new Vinyl Simulator sitting on its own button, and BOSS sold the machine on price, calling it "easily the most affordable phrase sampler on the market".

The specifications read like a constraint list. Eight pads and two banks, so sixteen sounds. Twenty-bit converters and three sample rates: 44.1 kHz for thirty-one seconds, 22.05 for sixty-three, and 11.025 kHz for three minutes and ten seconds of deliberately grainy audio. A SmartMedia slot for more. Sixteen patterns of sequencing, which people mostly ignored. No undo, no screen worth the name, and no way to fix a chop except to sample it again.

The Vinyl Simulator

Vinyl Simulator is the reason the used prices never fell. It stacks the artifacts of playing a record: a narrowed frequency range, added noise, and a compression that pushes the quiet parts of a loop up until the whole thing sits in one dense block. Run a clean break through it and the drums stop sounding like separate hits in a stereo field and start sounding like four seconds lifted off a forty-five.

The other half of the sound is the sampler itself. Pitching a sample up or down on the 303 introduced its own digital grit, and the low sample-rate modes band-limited everything through a converter that was never trying to be transparent. Resampling let you print all of it and run the result back through again, so a producer could stack two or three generations of degradation in a couple of minutes.

DRUGS is our free one-knob bus compressor, tuned by ear, aimed straight at that squashed, dense drum sound. It is not a model of any machine; it goes at the sound.

What the vinyl sim does to dynamics

The compression is the half of the effect nobody explains. Search traffic for sp 303 vinyl sim compression comes from producers who already got the filter and the noise right, ended up with something dull instead of dusty, and worked out on their own that the missing part lives in the level rather than the tone.

Here is the behavior. Peaks come down, the space between the hits comes up, and the distance between the two closes. That distance is crest factor, and narrowing it is why a processed loop reads louder on a phone while the peak meter has barely moved. A kick keeps its weight and loses a sliver of its leading edge. A snare keeps its body and loses the tip. The room tone that was hiding between hits climbs until it belongs to the loop instead of sitting under it.

Timing is what makes the squash read as a record rather than as a mix decision. The level drops on the hit and takes most of a bar to come back, so the recovery is audible as a swell behind the next two or three beats. A fast return gives you the pumping of a modern drum bus. A slow one gives you the sense that everything in the loop is riding in a single groove, rising and falling together, which is the thing people mean when they say a break sounds like it came off a forty-five.

It changes what your transients are for. Once the first few milliseconds of every hit have been rounded off, the ear stops using attack as its cue for loudness and starts using density. That is why a squashed break can sit several dB lower in a mix and still feel like the loudest thing in the track, and why turning the drum bus down after the fact does not undo the effect.

Where the squash in a vinyl sim comes from

A record is compressed several times before anyone hears it. Groove excursion is finite, so the cutting engineer rides level and limits ahead of the lathe, harder on the low end than anywhere else, and harder again on a seven-inch where the playing time per side is short. Playback adds its own share. A cartridge tracking a worn groove is not a linear device, and a portable player with a ceramic cartridge is nowhere near one.

Stack all of that and you get a signal whose loud parts have been pulled toward the middle and whose quiet parts have been pushed up to meet them. An effect labeled vinyl collapses the whole chain into one button. The good ones stay program dependent: more squash on a dense bar than a sparse one, more on the low end than the top, and a recovery that follows the material. The thin ones apply one fixed curve and hand you a low-pass with hiss on it.

It is also why the effect does more on a full loop than on separated tracks. Everything shares one path, so the kick is what ducks the hi-hats and the hi-hats are what fill the gap the kick leaves. Glue compression is the same mechanism described without the record.

DRUGS is our free one-knob bus compressor: the squash on its own, with no noise bed and no filter. It is not a model of any machine. It goes at the part most attempts leave out.

Who used it, and on what

Madlib took one to Red Bull Music Academy in São Paulo in 2002 with a portable turntable and a tape deck, and told Scratch magazine in 2005 that "I like to move quickly, and these little boxes are easy to use". Material from that trip fed into Madvillainy, released in 2004. MF DOOM, who rhymed over those beats, said he came to the 303 through Madlib and Jay Dee.

J Dilla got his through Stones Throw. Karriem Riggins has described what arrived: "They brought him a little BOSS [SP] 303 sampler and little 45 record player... That's what brought him through to make a lot of music that we hear on Donuts." The album came out on 7 February 2006, his thirty-second birthday, three days before he died. Accounts of exactly what else was in that room differ, and the people who visited at different points remember different machines, but a small sampler and a portable record player are in every telling.

That is the whole case for the machine, and it is worth being precise about it: the 303 did not make those records good. It made a particular kind of limitation cheap and portable at the moment two producers wanted exactly that.

How to get the sound now

1. Start with a full loop, not a multitrack: the sound depends on everything sharing one path.
2. Band-limit it. Roll the top off hard, and drop the sample rate if you want the grain.
3. Compress until the room and the noise come up behind the hits, not until it is louder.
4. Add a continuous noise bed rather than clicks and pops: the effect is a medium, not a fault.
5. Print it. Bounce the processed loop and work on the bounce, which is what resampling forced.
6. Do the whole thing again on the bounce if you want the second generation.
7. Keep it in mono, or close, and stop before it sounds broken.

If you want the crush and the filter part specifically, how to make a sample sound old covers it, and how to make drums sound dusty is the drum version. For the history of what came after this machine, see the SP-404 story.

Setting the compression, not just the tone

Do the level before you do anything to the tone. Nearly every attempt that comes out dull was built the other way round: filter, then noise, then a compressor bolted on at the end to rescue what was already wrong.

1. Route the full loop to one bus and put a compressor across it, before any filtering.
2. Set the attack long enough to let the first transient through, then shorten it until the kick's leading edge softens. On most breaks that lands between 3 and 15 ms.
3. Set the release long. You want the level recovering across most of a bar rather than between hits, which at 90 BPM is roughly 400 to 600 ms.
4. Drive it until gain reduction is moving 4 to 8 dB on the hits and never quite returns to zero inside the bar.
5. Match the output back to the level you started at, by ear, with the bypass.
6. Now band-limit. Roll the top off, and only then decide whether you also want the sample-rate drop.
7. Add the noise bed last and set its level against the compressed loop, since the squash has already raised everything the noise has to compete with.
8. Print it and work on the print.

The order carries most of the result. Compress first and the noise ends up living inside the signal, because the compressor is already reacting to the material the noise will share a floor with. Compress last and the noise gets squashed as a separate layer riding on top, which is the tell that people hear immediately and cannot name.

The costs are real. Four to eight dB of slow-release gain reduction makes a busy bar quieter than a sparse one, which is a groove change as much as a tone change, and a long cymbal decay will pump against the kick. Once you print, you own it.

How to tell whether you got the squash

Bounce two versions and play them on a phone. Not on the monitors, where the difference is easiest to talk yourself into.

On the version with the compression working, the loop stays the same distance from you all the way through the bar, and the quiet ticks between the backbeats come through a phone speaker that cannot reproduce the kick at all. On the version with only filtering and noise, the hits still poke and the space between them goes flat, because nothing lifted it.

The second test is the fader. Pull the drum bus down 6 dB inside the mix. A properly squashed loop is still legible down there. A merely band-limited one disappears, because all of its information was in the peaks, and the peaks were the first thing the filter took.

If the squash is too much of a good thing, blend rather than back off. Keeping a hard-squashed copy under the dry loop holds the attack and still lifts the room; parallel compression covers the method, and it is the standard fix when a break needs the density but cannot lose the snap.

Where it costs you

The constraints that made the records are real constraints. Sixteen sounds, no undo, and no editing to speak of means committing to a chop you cannot revise, and thirty-one seconds of full-rate audio is one break and a horn stab. Buying one now costs more than it did new, for a machine with aging rubber pads and a card format that ended twenty years ago. And the sound has become common enough that leaning on it is a genre signal rather than a decision.

The useful part of the lesson is the workflow, not the hardware. One signal path, one pass, print it, move on.

SP303, SP-303, and what people are actually searching for

The machine gets written both ways and the people writing it usually want one of three different things, which is worth separating before going further.

Some want the hardware — what it did, what it cost, whether it is worth buying one now. Some want one specific effect off it, the vinyl simulator, because they heard it on a record and could not name it. And some want to know how to get that sound today without buying a twenty-year-old sampler. The three questions have completely different answers and they get tangled together constantly.

The short version of the third one: the effect is a combination of things rather than a single process, and the combination is reproducible. What is hard to reproduce is not the filtering or the noise, which any plugin can do. It is the squash — the way the level collapses and comes back — and that is the part most attempts leave out.

The four parts of the vinyl sim sound

Pulled apart, it is four things happening at once, and skipping any of them is why a recreation sounds close but wrong.

Bandwidth. The top goes and some of the bottom goes with it. Not a steep filter — a narrowing at both ends that leaves the middle intact and slightly forward, which is why the result sounds closer rather than duller.

Noise. A bed underneath that is part of the signal path rather than laid over it. It shares the same top-end limit as the music, which is what stops it sounding like a separate layer.

The squash. The level collapses on the hit and returns across the bar. This is the part that makes drums sound like they came off a record, and it is the part a filter-and-noise recreation has none of.

Wear. The whole thing sounds like it has been played rather than pressed, which in practice means the top is not just quieter but less defined, and the stereo image is not quite as certain as it started.

Get the first two and you have a lo-fi filter. Get the third and it starts sounding like a record.

The effect on its own, separated from the machine, is in vinyl sim and what a vinyl sim plugin does.

The same effect name on the later machines

The name carried forward, which is where a lot of the confusion starts. Roland's own page for the SP-404MKII lists Vinyl Simulator among what it calls the SP classics, next to DJFX Looper, with Lo-fi, Cassette Simulator and Resonator listed as new, and separately advertises onboard side chain compression as a feature of its own.

Read that carefully, because it answers the question people keep typing as sp404 mk2 303 vinyl sim. The vinyl effect and the compressor are sold as two different things on that machine, and the feature list does not describe what either version of the vinyl effect does to level. A name on a button is a family resemblance, not a specification. Two machines a generation apart can carry the same label and behave differently in exactly the part that matters here.

The practical version, for anyone searching sp404 vinyl sim plugin or 303 vinyl sim plugin: choose for behavior, not for the name on the badge. Send a break through it and watch whether the level moves and how long it takes to come back. If the meter never budges, you have a filter with a noise generator attached, and the squash has to come from somewhere else. Software versions of the later machine covers the sampler workflow, and how to get an old sampler sound covers the tone underneath all of it.

Questions people ask

What is the SP-303?

A compact phrase sampler released by BOSS in 2001 as the successor to the SP-202. Eight pads, two banks, twenty-six effects including the Vinyl Simulator, internal resampling, and a SmartMedia card slot for extra storage.

What are the SP-303 effects?

Twenty-six in all. The ones people bought it for are Vinyl Simulator, Lo-Fi, Isolator, filter, distortion, Slicer and a compressor, with reverb, delay, flanger and a voice transformer filling out the list.

Does the SP-303 have a sequencer?

It has pattern recording: sixteen patterns, two banks of eight, holding up to 7,500 notes each. It is a phrase sequencer for arranging pads, not a full production sequencer, and most people who made records on it played the pads instead.

Who used the SP-303?

Madlib, on the sessions that fed Madvillainy, and J Dilla, who was given one with a 45 player by Stones Throw and used it on the music that became Donuts. MF DOOM said he came to the machine through both of them.

What is the vinyl simulator on the SP-303?

An effect that imitates the sound of a record: a narrower frequency range, added surface noise, and compression that lifts the quiet parts of a loop into one dense block. It is the single effect the machine is remembered for.

Is SP303 the same as SP-303?

Same machine, two spellings. It turns up both ways in forum posts, listings and video titles, which is why searching one and not the other misses half of what is written about it.

What does a vinyl sim actually do?

Four things at once: it narrows the bandwidth at both ends, adds a noise bed that shares the signal's own top-end limit, squashes the level so it collapses on hits and returns across the bar, and takes the definition off the top. The squash is the part most recreations leave out.

Does the SP-303 vinyl sim compress?

Yes, and that is the part of it people underestimate. Alongside the band-limiting and the noise, the effect pulls the loud parts down and lifts the quiet parts, so a break comes out denser at roughly the same peak level. Copy the filter and the hiss and skip this and you get a dull loop rather than a dusty one.

How do I get 303 vinyl sim compression without the noise?

Run a bus compressor with a slow release across the whole loop and leave the noise off. Attack long enough that the kick still reads, release long enough that the level recovers across most of a bar, and 4 to 8 dB of movement on the hits. That is the squash on its own, and on drums it carries most of the character.

Is the vinyl sim on the SP-404MKII the same as the 303 one?

Roland lists Vinyl Simulator on the MKII as one of the SP classics carried over, and lists side chain compression separately as its own onboard feature. Neither feature list states what either version does to level, so treat the shared name as a family resemblance and judge it by sending a loop through it.

Where should a vinyl sim go relative to my compressor?

If the sim has no dynamics of its own, compress ahead of it. If it does, use one or the other rather than both, or you will hear two recoveries fighting across the bar. Drum bus chain order covers the rest of the sequence.

What should a vinyl sim plugin be doing to my drum bus?

Four things: narrow the bandwidth at both ends, add a noise bed that shares the signal's own top limit, take the definition off the top, and move the level so it collapses on hits and returns across the bar. Most do the first three. The fourth is the one to check before you buy.

Why does my vinyl sim make the drums weaker?

Because you took the top off and put nothing back. Band-limiting alone removes the part of a hit that reads as attack, and if the level never moves, nothing rises to fill the space it left. Add the compression and the loop gets its weight back from the room instead of from the tips of the transients.

Is the vinyl sim squash the same as a normal drum bus compressor?

The mechanism is the same and the settings are not. A standard bus compressor is usually set to keep the hits intact and tidy the tails. The vinyl version is set long and deep enough that the loop breathes as one object across the bar. How to compress drums has the conventional starting points if you want to hear the difference back to back.

What DRUGS does

DRUGS is our free bus compressor with one macro control, called DOSE. Vinyl sim is shorthand for a particular kind of squash: dense, slightly crushed drums with the room pushed up under the hits, the way a break sounds once it has been through a sampler. DRUGS goes at that sound directly, on the drum bus, with one knob and no setup. Turning DOSE up makes the bus denser rather than louder, because peak level barely moves across the whole range of the knob, and the release reacts to the transient in front of it rather than asking you for a number.

Thirty-one seconds, eight pads, no undo. People are still trying to get back there.

Keep reading

• How to make a sample sound old — Narrow the range, add warmth and grain, wobble, noise and a small room.
• The SP-404 and the sound of Los Angeles — How a cheap sampler its makers underestimated became the instrument of LA's beat scene.
• MF DOOM — The date everyone had wrong, the method in his own words, and the machines.
• Madlib — The hotel-room rig, whole-record sampling, and ten minutes a beat.
• Kenny Segal — Live players over samples, the 404 as an effects box, and editing hit by hit.
• Free vinyl sim plugins, and the one we make — The free plugin for the noise, the free one for the squash, and the order.