| name | livepilot-sample-engine |
| description | This skill should be used when the user asks to "sample", "chop", "slice a loop", "find me a sample", "load a sample", "flip this", "resample", "vocal chop", "found sound", "texture from", "turn this into", "break", "one-shot", "load into Simpler", or when get_sample_opportunities finds gaps in the song's sample usage.
|
Sample Engine — AI Sample Manipulation for Ableton Live
The Sample Engine is LivePilot's intelligence layer for sample discovery,
analysis, critique, and creative manipulation. It turns raw samples into
musical elements through 29 techniques drawn from Dilla, Burial, Amon Tobin,
Four Tet, and DJ Premier.
Philosophy: Surgeon vs Alchemist
Every sample workflow is guided by one of two philosophies:
- Surgeon — Precision. Match key, align tempo, carve frequencies, blend
seamlessly. The sample should sound like it was always part of the track.
- Alchemist — Transformation. Reverse, stretch, destroy, rebuild.
The original creator shouldn't recognize their sample.
- Auto (default) — Context decides. Building a clean layer? Surgeon.
Stuck and need surprise? Alchemist. The critics and intent determine which.
6 MCP Tools
| Tool | Purpose |
|---|
analyze_sample | Build SampleProfile — material type, key, BPM, recommendations |
evaluate_sample_fit | 6-critic battery — key, tempo, frequency, role, vibe, intent fit |
search_samples | Search Splice library, Ableton browser, and local filesystem |
suggest_sample_technique | Recommend techniques from the 29-recipe library |
plan_sample_workflow | End-to-end plan: analyze + critique + technique + compiled steps |
get_sample_opportunities | Analyze song for where samples could improve it |
Workflow Modes
Direct Request
User asks to do something specific with a sample:
"Chop this vocal into a rhythm"
-> analyze_sample -> evaluate_sample_fit(intent="rhythm")
-> suggest_sample_technique -> execute
Discovery Mode
User wants to find and use a sample:
"Find me a dark vocal for this track"
-> search_samples(query="dark vocal") -> present candidates
-> user picks -> analyze -> critique -> plan -> execute
Wonder Mode Integration
When stuck, Wonder Mode can suggest sample-based variants:
- 6 sample-domain semantic moves in the registry
- Compiled plans require agent to supply
file_path from prior search_samples results
- Diagnosis detects: no_organic_texture, stale_drums, dense_but_static
- Note: Wonder does not yet auto-call
search_samples — the agent must chain the calls
Golden Rules
- Always analyze before loading —
analyze_sample tells you what
the material is before you commit to a technique
- Always critique before executing —
evaluate_sample_fit catches
key clashes, tempo mismatches, and frequency masking before they happen
- Respect the intent — "rhythm" and "texture" need different approaches
even for the same sample
- Start from the nearest technique — don't improvise a workflow when
a proven recipe exists in the library
- Present both plans — surgeon and alchemist. Let the user choose.
Material Types
| Type | Detection | Best Simpler Mode | Best Warp Mode |
|---|
| vocal | "vocal", "vox", "voice" in name | Slice (Region) | Complex Pro |
| drum_loop | "drum", "break", "beat" in name | Slice (Transient) | Beats |
| instrument_loop | "guitar", "piano", "synth" | Slice (Beat) | Complex Pro |
| one_shot | "kick", "snare", "clap", short | Classic | Complex |
| texture | "ambient", "pad", "drone" | Classic | Texture |
| foley | "foley", "field", "recording" | Classic | Texture |
| fx | "fx", "riser", "sweep" | Classic | Complex |
| full_mix | full mix, long duration | Slice (Beat) | Complex Pro |
6 Sample Critics
Each scores 0.0-1.0 on one dimension of fitness:
- Key Fit — Circle-of-fifths distance from song key
- Tempo Fit — BPM match including half/double time
- Frequency Fit — Spectral overlap with existing mix
- Role Fit — Does this fill a missing role in the song?
- Vibe Fit — Taste graph alignment (if evidence exists)
- Intent Fit — Does the material serve the stated goal?
Beyond Simpler — Sample-Adjacent Instruments in the Library
Simpler is the default sample container, but the installed packs ship instruments that change the physics of what playback means. When the user reaches for "more texture" / "less recognizable" / "evolving" / "grain" / "drone bed", consider these before defaulting to Simpler + warp:
- Granulator III (Live Suite, requires Live 12.0+ Max for Live) — Robert Henke's granular synth. Three modes: Classic (Granulator II engine), Loop (rhythmic playback), Cloud (up to 20 unsynced overlapping grains). MPE per-note control of grain size/shape/position. Built-in real-time audio capture — can grain whatever is currently playing in the session in one click. 59 presets. Reach for it when the user wants a sample to evolve, smear, or stop sounding like itself.
- PitchLoop89 (Live Suite) — Henke / Publison DHM 89 emulation. Two independent pitch-shifting delay voices, ±24 semitones, fine ±1 semitone, per-voice feedback / pan / reverse. The right tool when "shimmer", "detuned cloud", or "pitched echo" describes the goal.
- Drone Lab's Harmonic Drone Generator — 8-voice M4L synth (Expert Math). Multiple tuning systems including just intonation, Pythagorean, Indonesian Pelog. The right tool when the source isn't a sample at all but a sustained tonal bed under whatever's already playing.
- Lost and Found multisamples — music box, mid-century Capri fan organ, DIY plastic guitar, melodica, kazoo, metal xylophone, plus 7 Drum Racks built from everyday objects (steel water bottle, analog clock, submerged objects). When the user wants found-sound character without recording their own.
These are awareness, not prescription — pick the one that fits the moment. Don't lock in a chain.
Reference Docs
references/sample-techniques.md — Full 29-technique catalogreferences/sample-critics.md — Critic scoring detailsreferences/sample-philosophy.md — Surgeon vs Alchemist guidedocs/research/2026-04-22-ableton-packs-deep-analysis.md — Per-pack capabilities (44 installed packs scored against minimal/dub/cinematic aesthetic)
