The program builder for your vintage MPC.
The program builder for your vintage MPC.
The program builder for your vintage MPC.
Build kits and sequences in the browser, export to hardware.
Build kits and sequences in the browser, export to hardware.
Build kits and sequences in the browser, export to hardware.
RETRO2000 is built around the MPC2000 workflow: samples, chops, pads and sequences. Prepare your kit on a bigger screen, then take it back to the machine.
RETRO2000 is built around the MPC2000 workflow: samples, chops, pads and sequences. Prepare your kit on a bigger screen, then take it back to the machine.
Why RETRO2000
More room to work. Same destination.
More room to work. Same destination.
More room to work. Same destination.
The goal isn’t to replace your MPC. It’s to get your samples and programs ready so you can spend more time making music on it. Load your sounds, make the chops, set up the pads, shape and process them, and try a sequence. Export when you’re ready. Your hardware is still where the beat ends up.
The goal isn’t to replace your MPC. It’s to get your samples and programs ready so you can spend more time making music on it. Load your sounds, make the chops, set up the pads, shape and process them, and try a sequence. Export when you’re ready. Your hardware is still where the beat ends up.
Features
Everything in one window.
Everything in one window.
Everything in one window.
Each part of the program, as it ships.
Each part of the program, as it ships.
01 / SAMPLE EDITOR
Chop, trim and clean up samples.
Chop, trim and clean up samples.
Chop, trim and clean up samples.
Zoomable waveform with chop markers. Slice by hand, by threshold or by bars, and clean the edges with de-pop. Trim, normalize, transpose and warp the sample, then send the chops straight to pads.
Zoomable waveform with chop markers. Slice by hand, by threshold or by bars, and clean the edges with de-pop. Trim, normalize, transpose and warp the sample, then send the chops straight to pads.
02 / INPUT RECORDING
Record straight into the sample pool.
Record straight into the sample pool.
Record straight into the sample pool.
Choose an audio input, check the level on the input meter, then arm and record. Stop saves the take into your sample pool, ready to chop. Recordings are capped at two minutes. Your audio stays on your machine.
Choose an audio input, check the level on the input meter, then arm and record. Stop saves the take into your sample pool, ready to chop. Recordings are capped at two minutes. Your audio stays on your machine.
03 / PROCESS - COLOR
Color the sound.
Color the sound.
Color the sound.
Process opens four tabs. Color is the front door: drive modes, bit depth, sample rate, mix and output trim, with auto gain to keep levels honest. Flip A/B to hear processed against dry. Print the result as a new sample or print it to a pad.
Process opens four tabs. Color is the front door: drive modes, bit depth, sample rate, mix and output trim, with auto gain to keep levels honest. Flip A/B to hear processed against dry. Print the result as a new sample or print it to a pad.
04 / PROCESS - FX
Reverb, delay and modulation.
Reverb, delay and modulation.
Reverb, delay and modulation.
Reverb, delay, vibrato, phase, flanger and chorus, with low and high cuts in and out and a parallel wet control.
Reverb, delay, vibrato, phase, flanger and chorus, with low and high cuts in and out and a parallel wet control.
05 / PROCESS - STRIP
A console-inspired channel strip.
A console-inspired channel strip.
A console-inspired channel strip.
Filter, gate and transient shape, four-band EQ, compressor, drive and output trim on one strip. It is console-inspired, not a licensed emulation of any specific desk.
Filter, gate and transient shape, four-band EQ, compressor, drive and output trim on one strip. It is console-inspired, not a licensed emulation of any specific desk.
06 / PROCESS - HW FX
Run it through your own hardware.
Run it through your own hardware.
Run it through your own hardware.
Send a sample out through an output pair of your interface, record the return and print it as a new sample or replace the pad. Calibrate the loop first, with a test tone and a return level meter. Keep your speakers low the first time. The app guards against runaway levels on the return.
Send a sample out through an output pair of your interface, record the return and print it as a new sample or replace the pad. Calibrate the loop first, with a test tone and a return level meter. Keep your speakers low the first time. The app guards against runaway levels on the return.
07 / PARAMS
Shape every pad.
Shape every pad.
Shape every pad.
Attack, decay and mode. Filter frequency, resonance and envelope. Tune, level and pan. Velocity routing for level, attack, start, frequency and pitch. Play modes: mono, note on, mute groups, reverse. The preview is an approximation of the MPC2000’s envelope and filter. The exported program carries the exact values.
Attack, decay and mode. Filter frequency, resonance and envelope. Tune, level and pan. Velocity routing for level, attack, start, frequency and pitch. Play modes: mono, note on, mute groups, reverse. The preview is an approximation of the MPC2000’s envelope and filter. The exported program carries the exact values.
08 / MIXER
Balance the whole bank.
Balance the whole bank.
Balance the whole bank.
See the whole bank in one place and balance it before you export. Pan follows an equal-power law, as the 2000’s does: the centre sits 3 dB below a hard pan.
See the whole bank in one place and balance it before you export. Pan follows an equal-power law, as the 2000’s does: the centre sits 3 dB below a hard pan.
09 / MULTI OUT
Route pads to individual outputs.
Route pads to individual outputs.
Route pads to individual outputs.
Route pads to individual outputs and the routing is written into the exported program, so your MPC sends them to the right jacks. Honest limit: the browser plays stereo. Individual outs change the exported routing, not what you hear in the app.
Route pads to individual outputs and the routing is written into the exported program, so your MPC sends them to the right jacks. Honest limit: the browser plays stereo. Individual outs change the exported routing, not what you hear in the app.
10 / EB16 FX
EB16 effect sends.
EB16 effect sends.
EB16 effect sends.
Set per-pad sends for the EB16’s effects: two multi-effect buses and two reverb buses. Tick EB16 to hear a preview of the effects. Choose the effects themselves on the MPC (FXedit). The preview is a guide, not a replacement for your board.
Set per-pad sends for the EB16’s effects: two multi-effect buses and two reverb buses. Tick EB16 to hear a preview of the effects. Choose the effects themselves on the MPC (FXedit). The preview is a guide, not a replacement for your board.
11 / DUCK
Sidechain-style ducking.
Sidechain-style ducking.
Sidechain-style ducking.
Pick a source and a target, set depth, attack, hold and release, choose a linear or exponential shape and preview it. Apply to print the ducked result.
Pick a source and a target, set depth, attack, hold and release, choose a linear or exponential shape and preview it. Apply to print the ducked result.
12 / TRACKS
Keep a growing kit readable.
Keep a growing kit readable.
Group pads into named tracks so a sequence stays readable as the kit grows. Solo, mute and hide any track. Print tracks with an offset, and archive them as MIDI.
12 / TRACKS
Keep a growing kit readable.
Group pads into named tracks so a sequence stays readable as the kit grows. Solo, mute and hide any track. Print tracks with an offset, and archive them as MIDI.
13 / SEQUENCER
Record and overdub.
Record and overdub.
Record and overdub.
Four banks of sixteen pads, named and coloured by track. Note repeat, undo and erase work the way you expect. Record and overdub from your pads or a MIDI controller. Four sequence slots, tap tempo, count-in, click and timing correct. Save as a MIDI file, resample to the pool or bounce to audio.
Four banks of sixteen pads, named and coloured by track. Note repeat, undo and erase work the way you expect. Record and overdub from your pads or a MIDI controller. Four sequence slots, tap tempo, count-in, click and timing correct. Save as a MIDI file, resample to the pool or bounce to audio.
14 / STEP SEQUENCER
Build patterns step by step.
Build patterns step by step.
Sixteen steps per bar, with bar select, range select and duplicate. Pads stay above so you can audition while you build.
14 / STEP SEQUENCER
Build patterns step by step.
Sixteen steps per bar, with bar select, range select and duplicate. Pads stay above so you can audition while you build.
15 / TIMING AND SWING
Swing, measured against real machines.
Swing, measured against real machines.
Swing, measured against real machines.
Pick a timing grid and a swing percentage, then apply it to notes already in the sequence or hear it live. One swing model, measured against real machines. See the proof below.
Pick a timing grid and a swing percentage, then apply it to notes already in the sequence or hear it live. One swing model, measured against real machines. See the proof below.
16 / EXPORT AND BOUNCE
Take it back to the machine.
Take it back to the machine.
Download the program and sounds, transfer them to your MPC, and check the kit on the machine before relying on it for a session. Bounce the sequence to a WAV to hear it outside the browser.
16 / EXPORT AND BOUNCE
Take it back to the machine.
Download the program and sounds, transfer them to your MPC, and check the kit on the machine before relying on it for a session. Bounce the sequence to a WAV to hear it outside the browser.
HOW IT’S MODELED
Measured on real machines. Then checked by ear.
Measured on real machines. Then checked by ear.
Measured on real machines. Then checked by ear.
RETRO2000 does not guess what a vintage MPC does. We record the hardware, measure it, model it, then A/B the model against the machine and keep only what holds up. Here is what that looks like, with the numbers.
RETRO2000 does not guess what a vintage MPC does. We record the hardware, measure it, model it, then A/B the model against the machine and keep only what holds up. Here is what that looks like, with the numbers.
1 / Capture
Record the hardware.
Fixed test sounds, level ladders and ramps recorded from the real MPC into a clean interface, with loopback checks so the interface itself can be ruled out.
1 / Capture
Record the hardware.
Fixed test sounds, level ladders and ramps recorded from the real MPC into a clean interface, with loopback checks so the interface itself can be ruled out.
1 / Capture
Record the hardware.
Fixed test sounds, level ladders and ramps recorded from the real MPC into a clean interface, with loopback checks so the interface itself can be ruled out.
2 / Measure
Pull out the numbers.
Level, tone, distortion, pan law and timing, read from the recordings, not from the manual.
2 / Measure
Pull out the numbers.
Level, tone, distortion, pan law and timing, read from the recordings, not from the manual.
2 / Measure
Pull out the numbers.
Level, tone, distortion, pan law and timing, read from the recordings, not from the manual.
3 / Model and A/B
Build it, then compare.
The app renders the same test material and gets measured against the hardware take. If it misses, we find out why.
3 / Model and A/B
Build it, then compare.
The app renders the same test material and gets measured against the hardware take. If it misses, we find out why.
3 / Model and A/B
Build it, then compare.
The app renders the same test material and gets measured against the hardware take. If it misses, we find out why.
4 / Ear check
A producer listens.
The final call is by ear, on the owner’s own machine. Claims that fail stay out of the product.
4 / Ear check
A producer listens.
The final call is by ear, on the owner’s own machine. Claims that fail stay out of the product.
4 / Ear check
A producer listens.
The final call is by ear, on the owner’s own machine. Claims that fail stay out of the product.
MPC2000 and the EB16
Measured on one classic MPC2000 with an EB16, recorded through an Apollo 16.
EB16 distortion: the transfer curve, measured.
With the EB16’s Gain turned up, the board soft-clips. We measured it with a 1 kHz ramp at five gain settings. Hardware peaks run about 8.5 dB over digital full scale, and the soft-clip ceiling sits near 8.4 dB over unity, with the knee hardening as Gain rises. At Gain 99 the output actually folds back: it turns down as the input rises, and dips and recovers more than once. There is only a hint of it at 85 and none at 75 and below. The 2000 OUT stage models this curve, and it was judged accurate by ear with levels indistinguishable from the hardware. The 2000’s swing is measured too: it truncates the tick count, floor(0.48 x (swing - 50)) at 1/16, so 56% gives 2 ticks and 58% gives 3. That rule is part of 2000 OUT.
MPC2000 and the EB16
Measured on one classic MPC2000 with an EB16, recorded through an Apollo 16.
EB16 distortion: the transfer curve, measured.
With the EB16’s Gain turned up, the board soft-clips. We measured it with a 1 kHz ramp at five gain settings. Hardware peaks run about 8.5 dB over digital full scale, and the soft-clip ceiling sits near 8.4 dB over unity, with the knee hardening as Gain rises. At Gain 99 the output actually folds back: it turns down as the input rises, and dips and recovers more than once. There is only a hint of it at 85 and none at 75 and below. The 2000 OUT stage models this curve, and it was judged accurate by ear with levels indistinguishable from the hardware. The 2000’s swing is measured too: it truncates the tick count, floor(0.48 x (swing - 50)) at 1/16, so 56% gives 2 ticks and 58% gives 3. That rule is part of 2000 OUT.
MPC2000 and the EB16
Measured on one classic MPC2000 with an EB16, recorded through an Apollo 16.
EB16 distortion: the transfer curve, measured.
With the EB16’s Gain turned up, the board soft-clips. We measured it with a 1 kHz ramp at five gain settings. Hardware peaks run about 8.5 dB over digital full scale, and the soft-clip ceiling sits near 8.4 dB over unity, with the knee hardening as Gain rises. At Gain 99 the output actually folds back: it turns down as the input rises, and dips and recovers more than once. There is only a hint of it at 85 and none at 75 and below. The 2000 OUT stage models this curve, and it was judged accurate by ear with levels indistinguishable from the hardware. The 2000’s swing is measured too: it truncates the tick count, floor(0.48 x (swing - 50)) at 1/16, so 56% gives 2 ticks and 58% gives 3. That rule is part of 2000 OUT.
Artwork’s MPC3000
Artwork’s MPC3000
Artwork’s MPC3000
Measured on one MPC3000 running OS 3.11, owned and tested by Arthur, known as Artwork.
Measured on one MPC3000 running OS 3.11, owned and tested by Arthur, known as Artwork.
We recorded one rimshot with swing walked up from 52% to 75%. Swing moves in two-percent steps, one tick (at 96 PPQ) per step. A second, fresh sweep at 54, 58, 62, 66 and 71 percent at both 1/16 and 1/8, each applied from straight, matched RETRO2000’s standard swing to the tick in all 12 takes. Artwork A/B’d the result against his own machine and told us standard swing was the one. The 3000 output stage was measured too: velocity behaves as pure linear gain, second-harmonic distortion sits near -76 dBc and does not change with level, and the top end rolls off gently above 4 kHz.
We recorded one rimshot with swing walked up from 52% to 75%. Swing moves in two-percent steps, one tick (at 96 PPQ) per step. A second, fresh sweep at 54, 58, 62, 66 and 71 percent at both 1/16 and 1/8, each applied from straight, matched RETRO2000’s standard swing to the tick in all 12 takes. Artwork A/B’d the result against his own machine and told us standard swing was the one. The 3000 output stage was measured too: velocity behaves as pure linear gain, second-harmonic distortion sits near -76 dBc and does not change with level, and the top end rolls off gently above 4 kHz.
In progress
The 3000 model is still being finished. Swing and the output character are measured on one real machine. Other behaviours, including mono retrigger, note-off handling and loop-wrap tails, are still being tested. We will say so on this page when they are confirmed, and not before.
In progress
The 3000 model is still being finished. Swing and the output character are measured on one real machine. Other behaviours, including mono retrigger, note-off handling and loop-wrap tails, are still being tested. We will say so on this page when they are confirmed, and not before.
MIDI and audio interfaces
Plug in what you already own.
Plug in what you already own.
Plug in what you already own.
Controllers
Controllers
Controllers
Play pads from any class-compliant MIDI controller. PAD MAP learn assigns a note to each pad, so a keyboard or pad controller works as well as a standard map.
Play pads from any class-compliant MIDI controller. PAD MAP learn assigns a note to each pad, so a keyboard or pad controller works as well as a standard map.
Your MPC as the controller
Your MPC as the controller
Your MPC as the controller
Use an MPC2000 as a MIDI controller into the app, even with its scrambled factory note table. Learn the pad notes once and play the kit from the hardware.
Use an MPC2000 as a MIDI controller into the app, even with its scrambled factory note table. Learn the pad notes once and play the kit from the hardware.
Launchpad track control
Launchpad track control
Launchpad track control
Learn a Launchpad Mini (MK3, Programmer mode) onto track cells, with Mute, Solo, Rec and Hide modes. MIDI LEARN also maps knobs and faders to the process controls.
Learn a Launchpad Mini (MK3, Programmer mode) onto track cells, with Mute, Solo, Rec and Hide modes. MIDI LEARN also maps knobs and faders to the process controls.
Audio inputs
Audio inputs
Audio inputs
Pick any input your browser can see. FIND INPUTS rescans, and the input meter shows level before you record.
Pick any input your browser can see. FIND INPUTS rescans, and the input meter shows level before you record.
Hardware FX loop
Hardware FX loop
Hardware FX loop
Send out of an interface output pair, record the return, calibrate the loop and print it. Built and measured with an Apollo 16, and any interface with enough outputs can do the same.
Send out of an interface output pair, record the return, calibrate the loop and print it. Built and measured with an Apollo 16, and any interface with enough outputs can do the same.
MIDI clock
MIDI clock
MIDI clock
The app can follow incoming MIDI clock and start / stop. It does not send clock, and it has not been clock-tested against a real 2000 yet.
The app can follow incoming MIDI clock and start / stop. It does not send clock, and it has not been clock-tested against a real 2000 yet.
Honest limits: MIDI controllers need desktop Chrome or Edge. Browsers on iPad cannot see MIDI devices, so on an iPad you run the app on a computer and use the iPad as the remote. The browser plays stereo, so Individual Outs change the exported routing, not what you hear in the app. Bluetooth audio adds 100 to 300 ms of delay.
Honest limits: MIDI controllers need desktop Chrome or Edge. Browsers on iPad cannot see MIDI devices, so on an iPad you run the app on a computer and use the iPad as the remote. The browser plays stereo, so Individual Outs change the exported routing, not what you hear in the app. Bluetooth audio adds 100 to 300 ms of delay.
Hardware compatibility
Built for the hardware. Clear about the limits.
Built for the hardware. Clear about the limits.
Built for the hardware. Clear about the limits.
Tested
MPC2000
MPC2000
MPC2000
Program and sound export in the classic PGM / SND format. Tested on real hardware. 2000XL programs import and export back, but the XL itself is not hardware-tested.
Program and sound export in the classic PGM / SND format. Tested on real hardware. 2000XL programs import and export back, but the XL itself is not hardware-tested.
Tested
MPC1000
MPC1000
MPC1000
PGM / WAV export tested on real hardware. Tell us your OS when reporting a problem.
PGM / WAV export tested on real hardware. Tell us your OS when reporting a problem.
Tested
MPC3000
MPC3000
MPC3000
PGM / SND export tested on OS 3.11. Set the loader’s DRUM tracks from MIDI channel to 1 and Arrange tracks to AS IN FILE. Other OS versions not yet verified.
PGM / SND export tested on OS 3.11. Set the loader’s DRUM tracks from MIDI channel to 1 and Arrange tracks to AS IN FILE. Other OS versions not yet verified.
Tested by one owner
MPC4000
MPC4000
MPC4000
A RETRO2000 MPC2000-format export loaded on a real MPC4000 with pan, tune and volume intact. Confirmed by one beta tester on 1 October. It previously took Windows XP and extra tools. Sequences are not verified yet.
A RETRO2000 MPC2000-format export loaded on a real MPC4000 with pan, tune and volume intact. Confirmed by one beta tester on 1 October. It previously took Windows XP and extra tools. Sequences are not verified yet.
Not verified
MPC2500
MPC2500
MPC2500
Uses the MPC1000-format PGM / WAV path. Still needs hardware verification.
Uses the MPC1000-format PGM / WAV path. Still needs hardware verification.
Needs testing
MPC5000
MPC5000
MPC5000
An export target is present but needs real-machine testing. Owners wanted.
An export target is present but needs real-machine testing. Owners wanted.
Hardware formats have their own limits. Don’t expect every browser effect, routing choice or setting to transfer unchanged to every model. Check the exported program on your own MPC.
Hardware formats have their own limits. Don’t expect every browser effect, routing choice or setting to transfer unchanged to every model. Check the exported program on your own MPC.
Requirements and current limits
A browser, your samples and your MPC.
A browser, your samples and your MPC.
A browser, your samples and your MPC.
RETRO2000 runs in the browser. Best in desktop Chrome or Edge. MIDI controllers need Chrome or Edge. Bring your own samples; a larger screen and a keyboard give you room to use the full workflow.
RETRO2000 runs in the browser. Best in desktop Chrome or Edge. MIDI controllers need Chrome or Edge. Bring your own samples; a larger screen and a keyboard give you room to use the full workflow.
You’ll need a way to transfer the exported files to your MPC. RETRO2000 doesn’t send them directly to the machine.
You’ll need a way to transfer the exported files to your MPC. RETRO2000 doesn’t send them directly to the machine.
This is not a DAW plugin, hardware emulator, desktop installer or App Store app. RETRO2000 does not save your session: export your work before refreshing or closing the tab.
This is not a DAW plugin, hardware emulator, desktop installer or App Store app. RETRO2000 does not save your session: export your work before refreshing or closing the tab.
Activation
One buyer. Two machines. Offline after activation.
One buyer. Two machines. Offline after activation.
One buyer. Two machines. Offline after activation.
Enter the license key from your Gumroad receipt. Each machine activates online once, then works offline.
Enter the license key from your Gumroad receipt. Each machine activates online once, then works offline.
FAQ
Straight answers.
Straight answers.
Straight answers.
Get RETRO2000
Get RETRO2000
Get RETRO2000
One-time purchase. No subscription.
One-time purchase. No subscription.
Launch pricing is announced with the release.
Launch pricing is announced with the release.
Check your setup before buying. Launch policy is no refunds, with exceptions considered case by case. Please read the compatibility notes and requirements first. This does not remove any rights you have under applicable law or prevent Gumroad from issuing a refund under its own policies.
Beta
Help test it on real machines.
Help test it on real machines.
Help test it on real machines.
New beta acceptances are for MPC4000 and MPC5000 owners only. Signups stay open to everyone for launch details. Applying doesn’t guarantee beta access.
New beta acceptances are for MPC4000 and MPC5000 owners only. Signups stay open to everyone for launch details. Applying doesn’t guarantee beta access.
RETRO2000 is an independent project from RETRO GLEAM, built around the Akai MPC2000 workflow. It is not affiliated with, endorsed by or sponsored by Akai Professional. MPC model names identify the intended hardware formats. Trademarks belong to their respective owners. (c) 2026 Retro Gleam.