Extractors are the nodes that pull a specific feature out of a sound and let you treat it as its own object — the move that makes K2K a spectral sampler rather than just a graph editor. Every extractor outputs both the extracted layer and the residual (what’s left after extraction), so a single node turns one buffer into two parallel streams you can process independently. Transient Extractor isolates the percussive attacks, leaving everything else in the residual. HPSS Extractor separates harmonic content from percussive content. Formant Extractor pulls out vocal/resonance formants. Spectral Envelope Extractor separates timbre from excitation. Spectral Peak Extractor keeps just the strongest resonant peaks. Pitch Class Extractor extracts all instances of a chosen note class across all octaves. Noise Floor, Silence, Micro Transient, and Temporal Envelope round out the set, each isolating a different kind of feature. Pair every extractor with the matching scope color mode upstream so you can see what you’re about to extract before committing to it. This is the category that gets its own tutorial — Tutorial 03 in the Editor track — because everything K2K does that other samplers can’t begins here.
Nodes
- HPSS Extractor — HPSS source separation - extracts Harmonic, Percussive, or Residual component
- Spectral Envelope Extractor — Cepstral envelope/excitation separation - extract timbre or pitch
- Transient Extractor — Extract attack portions (first 10-50ms) of audio events using onset detection
- Noise Floor Extractor — Extract continuous noise floor (hiss, breath, room tone) using HPSS residual
- Micro Transient Extractor — Extract sub-10ms events (clicks, pops, vinyl crackle) using spectral flux
- Pitch Class Extractor — Extract musical notes/chords across all octaves using pitch class mapping
- Temporal Envelope Extractor — Extract per-frequency amplitude envelopes (smooth temporal curves)
- Silence Extractor — Extract quiet/silent regions based on magnitude threshold
- Spectral Peak Extractor — Extract resonant peaks (formants, overtones) from spectrum
- Formant Extractor — Extract vocal formants (F1-F4) using LPC analysis
Generated 2026-05-05 from K2K_Dev@96730bdc by scripts/gen_lexique.py. Edit _intros/ or _overrides/, not this file.