Samples are recordings. Fixed length, fixed pitch, fixed order. Three methods turn them into something else: speed changes pitch, chop cuts them into pieces, slice rearranges those pieces. Old audio becomes new material.
Repertoire: The Amen Break (The Winstons, 1969) — the 6-second drum solo that built half of electronic music. Dark World Theme (Koji Kondo, Zelda: A Link to the Past, 1991) — the moment game music moved from synthesis to samples. Corridors of Time (Yasunori Mitsuda, Chrono Trigger, 1995) — one of the most remixed game tracks ever.
Parallel composition: Your ambient piece gets texture. The clean synthesized koto becomes chopped, grained, material. The finished piece becomes more finished.
This is the last lesson of Phase 1. The piece ships.
Same clap sample. Four different playback speeds:
First hit: normal. Second: double speed, pitched up an octave. Third: half speed, an octave down. Fourth: reversed. Same sample, four characters.
The phonk aesthetic is about slowing everything down. Pitched-down samples, heavy reverb:
Everything darker. The sample is longer because it’s slower. The reverb smears it further. That’s the trick—speed down, reverb up.
Playback speed = pitch. .speed(2) = octave up, double speed. .speed(0.5) = octave down, half speed. .speed(-1) = reverse at normal pitch. Fractional values like 0.7 pitch-shift without landing on a clean interval.
Sixteen hi-hats, but some of them vanish:
Each hit has a 50% chance of playing. The pattern is different every cycle. ? after a pattern element means “maybe.” Add a number for specific odds: ?0.3 means 30% chance each hit plays. The rest are silence.
A different kind of randomness. One note per event, chosen from a set:
Four notes. Each cycle, one is picked at random. The pipe character | means “or.” Not a sequence—a coin flip between options.
Probability inside the pattern itself. ? drops events randomly (50% default, ?0.3 for 30%). | picks one option at random per event. Both make the pattern non-deterministic—different every time.
Try s("hh*16?0.2") for sparse dropouts—only 20% of hits survive. Or build a randomized kit: s("<bd|cp> <sd|hh> <bd|cp> <sd|oh>"). Each slot picks between two sounds every cycle.
A breakbeat sample, stretched to fit exactly 2 cycles:
.loopAt(2) adjusts playback speed so the sample spans exactly 2 cycles. The break loops in time with the tempo. No manual speed calculation.
Same break, but cut into 8 equal pieces:
Eight slices, played in order. The break sounds the same but it’s in pieces now. Each slice is a separate event. That means you can apply effects to individual slices, or rearrange them.
Instead of the whole sample, play only the middle quarter:
.begin(0.25) starts playback at 25% through the sample. .end(0.5) stops at 50%. Only that slice plays. Useful for isolating a specific hit or phrase from a longer recording.
.chop(N) slices a sample into N sequential pieces. .loopAt(N) fits a sample to N cycles by adjusting speed. .begin() and .end() set start and stop points as fractions (0–1). Three ways to control which part of a sample you hear.
Try .chop(16) for finer slices—more granular, more stutter. .chop(4) for chunkier pieces. Replace .chop(8) with .striate(8)—striate interleaves the slices instead of playing them sequentially. The rhythm rearranges itself.
Chopping keeps the original order. Slicing lets you pick the order yourself:
Eight slices, numbered 0 through 7. Played in a shuffled order: 0, 3, 2, 7, 1, 5, 4, 6. The break is rearranged. New rhythm from old material. Same sounds, different sequence.
Notice the pitch shifts. When slices are reordered, some play faster or slower to fit their new position. .splice() corrects for that:
Same slice order. But .splice() adjusts pitch so each slice sounds like it did in the original. The character stays. The order changes.
Stack it with the phonk treatment. Pitched down, reverbed, rearranged:
Slower. Darker. The reverb glues the rearranged slices together. This is the workflow: take a break, slice it, reorder, pitch down, add space.
.slice(N, pattern) cuts a sample into N slices and plays them in the order given by the pattern. .splice(N, pattern) does the same but with pitch correction—slices maintain their original pitch regardless of position. Rearranging slices creates new rhythms from existing material.
Reverse some slices with speed: .slice(8, "0 3 2 7").speed("1 1 -1 1"). The third slice plays backward. Try euclidean slice patterns: .slice(8, "0(3,8)")—three hits of slice 0, euclidean-spaced across 8 slots.
The Winstons. “Amen, Brother.” 1969. The B-side. A 6-second drum solo by Gregory Coleman between the second chorus and the outro. That’s it.
Those six seconds became the backbone of jungle, drum and bass, hip-hop, breakcore, and hundreds of other genres. The most sampled break in recorded music. It appeared on thousands of tracks before anyone thought to ask who played it.
Coleman died homeless in 2006. He never received royalties for the break. The Winstons never sued anyone over the sample—Richard Spencer, the group’s leader, said he didn’t believe in suing artists. The break entered public consciousness as if it had always been there. Free material. No credit, no payment, no recognition for decades.
In 2015, a GoFundMe raised money for a plaque honoring Coleman. The conversation about sampling ethics started late.
A basic amen-style break. Kick, snare, hats—the skeleton of the original:
And chopped. Sliced, rearranged, pitched down—the way it sounds in a jungle track:
Sampling is legal when cleared (licensed from the rights holder). Uncleared sampling is copyright infringement. The culture built itself on both. The tension between creative reuse and fair compensation is unresolved. The amen break is the clearest case: infinite creative output, zero compensation for the creator.
When you chop a sample in strudel, you’re working with the built-in library (Creative Commons / public domain). The ethics question applies when you load your own samples from copyrighted recordings.
E-mu SP-1200. Released 1987. 12-bit sampling. 26.04 kHz sample rate. 10 seconds of total sample time across all pads. Ten seconds.
Those constraints forced a sound. Short chops. Aggressive trimming. No room for long phrases—you took the hit you needed and nothing more. The low sample rate and 12-bit depth added grit. Frequencies above 13 kHz were gone. The sound was dark, crunchy, lo-fi.
Memphis, early 1990s. DJ Paul built the sound of Memphis rap on the SP-1200 and later the MPC. Chopped soul samples pitched down. Repetitive, hypnotic loops. The lo-fi quality wasn’t a limitation—it was the aesthetic. Dark, slow, heavy. The direct ancestor of modern phonk.
The SP-1200’s 10-second limit meant you couldn’t sample a whole bar of music. You sampled a single chord hit, a single drum, a vocal stab. Then you sequenced those tiny pieces into patterns. The repetition wasn’t laziness—it was the only option. And it created a hypnotic quality that became the genre’s signature.
.begin() and .end() are your SP-1200 constraints. .chop() automates what producers did manually for hours. The workflow is the same. The tools are faster.
Everything you played in L0-L6 was synthesized. Waveforms generated in code. Now listen to what happened when game composers got access to recordings.
Zelda: A Link to the Past — Dark World Theme (Koji Kondo, 1991). The SNES replaced the NES’s 5 synthesis channels with 8 sampled channels on the Sony SPC700 chip. Kondo went from triangle waves to sampled brass and choir. Same composer. Different material. Listen to how much richer it sounds:
Same notes. Different source material. The square wave is abstract. The trumpet sample is physical. That transition from synthesis to samples is what this lesson is about.
Chrono Trigger — Corridors of Time (Mitsuda, 1995). One of the most remixed tracks in game music history. People chop it, loop it, layer it. The SPC700 samples have a specific lo-fi quality that modern producers chase on purpose:
NES (1983): 5 channels of synthesis. SNES (1990): 8 channels of sampled audio on the Sony SPC700 chip. The chip could play back recordings, not just generate waveforms. Composers could record a trumpet, a violin, a choir, and play them back pitched to different notes.
NES composers were limited by what sounds they could make. SNES composers were limited by how much sample memory they had. The SPC700 had 64KB of audio RAM. A single second of CD-quality audio is 176KB. So every sample had to be tiny: short, looped, compressed. That compression gave SNES music its distinctive character. The lo-fi quality wasn’t a failure. It was the sound of fitting an orchestra into 64 kilobytes.
| system | audio | constraint | result |
|---|---|---|---|
| NES | 5 synth channels | what sounds you can make | inventive synthesis (Tanaka, Kondo) |
| SNES | 8 sample channels, 64KB RAM | how much memory you have | compressed samples, lo-fi warmth (Mitsuda, Uematsu) |
| PS1 | 24 channels, CD audio | almost none | full orchestral scores (Uematsu FF7) |
The lo-fi hip hop aesthetic that dominates YouTube streams in 2024? That’s people chasing the SPC700 sound on purpose. The constraint became the goal.
1986-91: UK producers discover Ultimate Breaks and Beats compilations. They sample drum breaks (especially the Amen) on Akai S950 samplers with 8 seconds of memory. Chop, pitch, loop. Jungle is born.
1994-97: J Dilla buys an Akai MPC3000. Instead of quantizing his samples to the grid, he leaves them slightly off. The timing is imperfect. It feels human. “Dilla time” becomes a production philosophy: the machine breathes like a person.
2000s: Nujabes (Japanese producer Seba Jun) blends jazz samples with Dilla’s timing, adds anime aesthetics. The proto-lo-fi hip hop sound. His soundtracks for Samurai Champloo bring Japanese and American sample culture together.
2010s-now: Lo-fi hip hop YouTube streams. The “study girl” aesthetic. The sound descends directly from SNES sample compression + Dilla timing + Nujabes jazz. Game music, chopped breaks, and Japanese production philosophy, all in one lineage.
Nujabes (1974-2010) is the connection between everything in this course. Japanese producer. Sampled American jazz. Used Dilla’s timing. Scored anime. His work on Samurai Champloo soundtracks fused traditional Japanese aesthetics with hip-hop production in a way nobody had done before. The lo-fi hip hop genre is essentially “what if we all tried to sound like Nujabes.”
Your piece is structurally complete from L6. Now add texture. The clean synthesized koto gets replaced with chopped, grained material. Same notes, different feel:
The koto is chopped into fragments. In the kyū section, .slice(8, "7 5 3 1 6 4 2 0") rearranges the fragments in reverse order. The final section degrades the chopped koto until only fragments remain, dissolving into reverb.
This is the piece. Seven lessons. From three notes in silence to a structured, evolving, textured ambient composition. Press share.
A phonk track with a chopped break at its center. Sliced, rearranged, pitched down, reverbed. The break alternates two different slice orders across cycles. Around it: 808 kick, clap, degraded hats, cowbell with delay, sub bass, and a filtered pad. Old recordings turned into new music.
"<[7 5 3 1] [6 4 2 0]>" for reversed slices—the break plays back to front..speed("<0.8 0.9 1 0.7>") to the chopped break for per-cycle pitch variation. Each cycle gets a different weight.break:1 with break:0 or break:2 for a different source break. Same slice pattern, different material.| tool | does | looks like |
|---|---|---|
| .speed(N) | playback speed = pitch shift | .speed(0.7) |
| .chop(N) | slice sample into N sequential pieces | .chop(8) |
| .slice(N, pat) | slice into N, play in pattern order | .slice(8, "0 3 2 7") |
| .splice(N, pat) | slice + pitch correction | .splice(8, "0 3 2 7") |
| .striate(N) | interleaved slicing | .striate(8) |
| .loopAt(N) | fit sample to N cycles | .loopAt(2) |
| .begin(N) | start playback at fraction N | .begin(0.25) |
| .end(N) | stop playback at fraction N | .end(0.5) |
| ? | degrade: random dropout (50%) | s("hh*16?") |
| ?N | degrade with probability N | s("hh*16?0.3") |
| | | random choice between options | note("c3|eb3|g3") |
.speed() for pitch shifting. .loopAt() to fit samples to the tempo. .chop() for sequential slicing. .slice() and .splice() for rearranging. .begin()/.end() for manual sample trimming. ? and | for probability in the pattern itself.
Next: The Scale. Numbers become notes. Melody gets a map.
Tracks that demonstrate this lesson’s concepts.
| artist | track | why |
|---|---|---|
| Koji Kondo | Zelda LttP: Dark World (1991) | (game) the SNES sample revolution. Synthesis → recordings. |
| Yasunori Mitsuda | Chrono Trigger: Corridors of Time (1995) | (VGM) the most remixed game track in history |
| Nujabes | Samurai Champloo: Battlecry (2004) | (anime/hip-hop) Japanese production + jazz samples + Dilla timing |
| DJ Shadow | Building Steam with a Grain of Salt (1996) | (hip-hop) first album made entirely from samples |
| J Dilla | Workinonit (2006) | (instrumental hip-hop) king of chopping, SP-303, “Dilla time” |
| Madvillain | Accordion (2004) | (hip-hop) pitched-down chop creates new identity |
| Pierre Schaeffer | Étude aux chemins de fer (1948) | (hip-hop) train recordings cut and rearranged: the first sample-based composition |
Before synthesizers, electronic music was made by cutting tape. The sampler returned to that idea—take a recording, cut it up, rearrange it—but at the speed of performance.
Pierre Schaeffer at RTF Paris used turntables and tape recorders to manipulate recorded sounds—train whistles, piano notes, spinning lids. He called it musique concrète: music made from “concrete” (recorded) sounds rather than “abstract” (notated) ones. “Étude aux chemins de fer” (1948) is built entirely from train recordings, cut and rearranged. This is chopping—45 years before jungle.
Dave Rossum’s E-mu SP-1200 sampled at 12-bit/26kHz with a maximum of 10 seconds of sample memory. The constraints were severe—you could only store short chops. Producers had to be surgical: grab just the kick, just the snare, just 2 seconds of a horn stab. The gritty, lo-fi character wasn’t an effect—it was the converter. Marley Marl, DJ Premier, Pete Rock, J Dilla, and the entire Native Tongues lineage produced on SP-1200s and its successors.
James Yancey (J Dilla) used Akai MPC3000 and Boss SP-303 samplers to develop a chopping style so distinctive it changed production permanently. His “drunk” timing—events intentionally off the quantize grid—humanized machine music. Donuts (2006), produced on the SP-303 from a hospital bed in his final weeks, is 31 tracks of chopped soul and jazz 45s, widely regarded as one of the greatest instrumental hip-hop albums.
The Winstons never copyrighted the amen break separately from the full song. When Richard Spencer (The Winstons’ frontman) tried to claim royalties in the 2000s, the legal landscape had already shifted—thousands of tracks had used it without clearance. In 2015, Spencer set up a GoFundMe for Gregory Coleman’s family. The amen break remains unprotected, a de facto public domain artifact that shaped multiple genres.
Sources: Schaeffer, In Search of a Concrete Music (1952/2012); Schloss, Making Beats (2004); Charnas, Dilla Time (2022); Harrison, “Can I Get An Amen?” (2004).