k2k audio Stretch, squeeze, reverse, freeze, scatter. These nodes reshape when things happen in your audio — from extreme slow-motion drones to rhythmic glitch stutters.
Time Reverse
What it does — Flips your audio backward in time, with the option to reverse only the magnitude, only the phase, or both together.
When you’d reach for it — You want a classic reversed cymbal swell, a reversed vocal phrase for a transition, or you’re after the uncanny texture that comes from reversing just the magnitude while leaving the phase intact.
Quick example
- Feed a drum loop into Time Reverse.
- Set the mode to Both for a standard full reversal.
- Listen to the result — cymbals now swell up into hits.
- Switch to Magnitude to keep the original timing feel but with a strange, smeared quality.
Parameters
| Parameter | What it controls | Options | Sweet spot hint |
|---|---|---|---|
| Mode | Which components get reversed | Magnitude / Phase / Both | Both for classic reverse; Magnitude for experimental texture |
Time Crop
What it does — Extracts a time region from your audio with adjustable fade-in and fade-out handles.
When you’d reach for it — You need to isolate a specific moment from a longer recording — a single phrase, a particular hit, or a slice of ambience — with clean, click-free edges.
Quick example
- Feed a full track into Time Crop.
- Set Start to 2.500 s and End to 5.000 s to grab a 2.5-second section.
- Add a 50 ms Fade In and 100 ms Fade Out with S-Curve for smooth edges.
- The output is just that cropped region, ready for further processing.
Parameters
| Parameter | What it controls | Range | Sweet spot hint |
|---|---|---|---|
| Start | Beginning of the crop region | 0 s to input duration | Depends on your material |
| End | End of the crop region | 0 s to input duration | Set to input length to crop only the start |
| Fade In | Duration of the opening fade | 0 – 1 s | 0.010 – 0.050 s for click-free cuts |
| Fade Out | Duration of the closing fade | 0 – 1 s | 0.010 – 0.100 s |
| Curve | Shape of both fades | Linear / Expo / S-Curve | S-Curve sounds most natural |
Time Concat
What it does — Joins two spectral buffers end-to-end with a crossfade at the seam.
When you’d reach for it — You’ve processed two separate audio regions differently and now need to stitch them back together smoothly, or you’re building a longer piece from shorter segments.
Quick example
- Connect the first segment to Input A and the second to Input B.
- Set Crossfade to 0.050 s with Equal Pow curve for a seamless join.
- Leave Gap at 0 for a tight edit, or increase it to insert silence between the segments.
Parameters
| Parameter | What it controls | Range | Sweet spot hint |
|---|---|---|---|
| Crossfade | Length of the overlap blend between A and B | 0 – 1 s | 0.030 – 0.100 s for clean joins |
| Curve | Shape of the crossfade transition | Linear / Equal Pow / S-Curve | Equal Pow preserves loudness best |
| Gap | Silence inserted between A and B | 0 – 5 s | 0 for tight edits |
Time Quantize
What it does — Shifts detected events in time toward attractors or grid lines, with phase-coherent spectral shifting under the hood.
When you’d reach for it — You have a loosely-timed performance and want to tighten events together creatively — not strict tempo quantization, but magnetic clustering where strong hits pull weaker ones closer, or where a variable grid reshapes the rhythmic feel.
Quick example
- Feed a drum recording into Time Quantize.
- Set mode to Attractor — the loudest hits stay put, and nearby quieter hits drift toward them.
- Turn Strength to 0.60 for a subtle tightening that keeps the human feel.
- Raise Threshold toward 0 dB to only affect the most prominent events.
Parameters
| Parameter | What it controls | Range | Sweet spot hint |
|---|---|---|---|
| Mode | Quantization approach | Attractor / Var Grid / Combined | Attractor for organic clustering |
| Strength | How far events move toward their targets | 0 – 1 | 0.50 – 0.80 keeps things musical |
| Threshold | Minimum event energy to be affected | -60 – 0 dB | -20 dB catches most transients |
| Max Shift | Farthest an event can move | 0 – 500 ms | 50 – 100 ms for subtle correction |
| Radius | How far an attractor’s pull reaches (Attractor/Combined) | 10 – 500 ms | 100 ms |
| Falloff | How attraction weakens with distance (Attractor/Combined) | Linear / Expo / Gaussian | Gaussian for smooth pull |
| Grid Int. | Spacing between grid lines (Var Grid/Combined) | 10 – 2000 ms | 100 – 250 ms |
| Phase | Strategy for shifting spectral frames | Direct / Source / OLA / Vocoder | Direct for transparency |
| Xfade | Crossfade frames at shift boundaries (Direct/OLA only) | 1 – 16 frames | 4 |
Time Warp
What it does — Remaps time non-linearly using a speed envelope, so different parts of your audio can play faster, slower, or even freeze.
When you’d reach for it — You want bullet-time slow-downs at a dramatic moment, expressive rubato across a phrase, or a scrubbing back-and-forth effect that no simple speed change can achieve.
Quick example
- Feed a vocal phrase into Time Warp.
- Set mode to Relative and draw a speed envelope that dips to 0.3x during the climactic word.
- Enable Preserve Dur. to keep the total length unchanged — the rest of the phrase speeds up to compensate.
- Choose Vocoder phase strategy for the smoothest result.
Parameters
| Parameter | What it controls | Range | Sweet spot hint |
|---|---|---|---|
| Mode | How the envelope is interpreted | Relative / Absolute | Relative for speed curves; Absolute to scrub freely |
| Speed | Local playback speed multiplier (Relative mode) | 0.1x – 4.0x | 1.0x is normal; below 0.5x gets dreamy |
| Preserve Dur. | Scale output to match input length | On / Off | On when you need the result to fit a specific slot |
| Phase | Strategy for interpolating warped frames | Direct / Source / OLA / Vocoder | Vocoder for cleanest stretching |
| Xfade | Crossfade frames at warp boundaries (Direct/OLA only) | 1 – 16 frames | 4 |
Paulstretch
What it does — Stretches audio to extreme lengths (up to 50x and beyond) by randomizing phase while preserving spectral color, turning any sound into a smooth, evolving drone.
When you’d reach for it — You want to turn a 3-second vocal clip into a 2-minute ambient pad, create meditative textures from field recordings, or generate slowly-evolving soundscapes from virtually any source material.
Quick example
- Feed a short piano chord into Paulstretch.
- Set Stretch to 20x — a 3-second chord becomes a 1-minute drone.
- Increase Window Size to 800 ms for an extra-smooth, glassy result.
- Adjust Gain if the output is too quiet or too hot.
Parameters
| Parameter | What it controls | Range | Sweet spot hint |
|---|---|---|---|
| Stretch | Time-stretch ratio | 1x – 50x | 8x – 20x for lush pads; 30x+ for deep ambience |
| Window Size | Analysis window length | 100 – 2000 ms | 400 – 800 ms; larger = smoother but less detail |
| Gain | Output level adjustment | -12 – +12 dB | 0 dB; bump up if the stretch thins the sound |
Grain Cloud
What it does — Chops audio into tiny overlapping grains and recombines them with independent pitch, time, and randomness controls.
When you’d reach for it — You want to pitch-shift without changing duration, create shimmering cloud textures from a voice or instrument, or add organic micro-variation to a static loop.
Quick example
- Feed a vocal sample into Grain Cloud.
- Set Grain Size to 80 ms and Density to 2.5x for smooth coverage.
- Shift Pitch up +7 semitones for a perfect fifth harmony layer.
- Add 20 cents of Pitch Jitter and 10 ms of Time Jitter to break up any mechanical feel.
- Set Mix to 100% for full wet, or 50% to blend with the dry signal.
Parameters
| Parameter | What it controls | Range | Sweet spot hint |
|---|---|---|---|
| Density | Grain overlap coverage | 1.0x – 4.0x | 2.0x – 3.0x for full, rich clouds |
| Grain Size | Duration of each grain | 10 – 500 ms | 30 – 100 ms for pitched material; 100+ ms for pads |
| Pitch | Pitch shift in semitones | -24 – +24 st | 0 for time-only effects |
| Envelope | Window shape applied to each grain | Hanning / Gaussian / Triangle | Hanning is the safest default |
| Pitch Jitter | Random pitch variation per grain | 0 – 100 cents | 5 – 25 cents for subtle life |
| Amp Jitter | Random amplitude variation per grain | 0 – 50% | 5 – 15% for organic texture |
| Time Jitter | Random start offset per grain | 0 – 50 ms | 5 – 15 ms to avoid flanging |
| Mix | Wet/dry balance | 0 – 100% | 50% for blending; 100% for full granular |
| Gain | Output level adjustment | -12 – +12 dB | 0 dB |
Frame Hold
What it does — Captures a block of spectral frames and loops them for a chosen duration, creating a sustained freeze of one moment in the audio.
When you’d reach for it — You want to freeze a particular vowel sound into an infinite drone, sustain a single chord for longer than it naturally rings, or create evolving textures by randomly jumping between frozen moments.
Quick example
- Feed a choir recording into Frame Hold.
- Enable Auto-detect to lock onto the loudest moment automatically.
- Set Block Length to 50 frames and Duration to 8 s for a long sustain.
- Add a 0.5 s Fade In and 1.0 s Fade Out for smooth entry and exit.
- Set Rate to 0.5 Hz to jump to a new random block twice a second for evolving texture.
Parameters
| Parameter | What it controls | Range | Sweet spot hint |
|---|---|---|---|
| Auto-detect | Automatically find the loudest frame | On / Off | On for quick results |
| Position | Manual start time (when Auto-detect is off) | 0 s to input duration | Scrub to find the right moment |
| Frame | Manual start frame (when Auto-detect is off) | 0 to total frames | Alternative to Position |
| Block Length | Number of consecutive frames to loop | 1 – 200 frames | 30 – 100 for natural-sounding loops |
| Duration | Total output length | 0.1 – 60 s | As long as you need it |
| Fade In | Opening fade duration | 0 – 5 s | 0.05 – 0.50 s |
| Fade Out | Closing fade duration | 0 – 5 s | 0.10 – 1.00 s |
| Jitter | Random frame variation within the block | 0 – 100% | 10 – 30% to soften loop points |
| Rate | How often to jump to a new random block | 0 – 20 Hz | 0 for static freeze; 0.5 – 2 Hz for slow evolution |
Spectral Delay
What it does — Applies a different delay time to each frequency band, so bass and treble arrive at different moments.
When you’d reach for it — You want frequency dispersion effects — bass lingering behind the treble, or highs trailing off separately — creating depth and movement that a normal delay cannot achieve. Great for sound design, spatial textures, and creating “impossible” acoustic spaces.
Quick example
- Feed a percussive loop into Spectral Delay.
- Set Delay to 300 ms and Tilt to -0.50 so low frequencies delay more than highs.
- Add 0.30 Feedback for repeating tails.
- Set LFO Rate to 0.3 Hz with 40% LFO Depth so the delay times slowly undulate.
Parameters
| Parameter | What it controls | Range | Sweet spot hint |
|---|---|---|---|
| Delay | Base delay time | 0 – 2000 ms | 100 – 500 ms |
| Feedback | How much delayed signal feeds back in | 0 – 0.95 | 0.20 – 0.50 for audible repeats; above 0.80 gets wild |
| Tilt | Frequency weighting for delay amount | -1 to +1 | 0 = all bands equal; negative = bass delays more |
| Mix | Wet/dry balance | 0 – 100% | 30 – 60% |
| Gain | Output level | -24 – +12 dB | 0 dB |
| LFO Shape | Modulation waveform for delay time | Sine / Tri / Saw+ / Saw- / Sqr / S&H | Sine for smooth modulation |
| LFO Rate | Speed of delay modulation | 0.01 – 20 Hz | 0.1 – 1 Hz for gentle movement |
| LFO Depth | Intensity of delay modulation | 0 – 100% | 20 – 50% |
| LFO Tilt | Frequency weighting for LFO intensity | -1 to +1 | 0 for uniform modulation |
Spectral Echo
What it does — Creates multiple decaying copies of the spectral content at regular intervals, with high-frequency rolloff that darkens each successive repeat.
When you’d reach for it — You need depth and space in the spectral domain — trailing ghost textures, rhythmic echo patterns, or a reverb-like tail that darkens naturally over time.
Quick example
- Feed a vocal snippet into Spectral Echo.
- Set Delay to 250 ms and Count to 6 echoes.
- Set Decay to -4 dB per echo for a gradual fade.
- Turn up HF Rolloff to 3000 Hz so each repeat gets progressively warmer and darker.
Parameters
| Parameter | What it controls | Range | Sweet spot hint |
|---|---|---|---|
| Delay | Time between echoes | 10 – 2000 ms | 150 – 400 ms for rhythmic echoes |
| Count | Number of echo repetitions | 1 – 20 | 3 – 8 for most musical uses |
| Decay | Volume drop per echo | -20 – 0 dB | -4 to -8 dB for natural fading |
| HF Rolloff | Frequency above which echoes decay faster | 0 – 10000 Hz | 2000 – 5000 Hz for warm, natural tails; 0 = no rolloff |
| Mix | Wet/dry balance | 0 – 100% | 30 – 60% |
| Gain | Output level | -24 – +12 dB | 0 dB |
Spectral Freeze Stutter
What it does — Freezes the current spectral frame whenever an internal LFO crosses a threshold, creating rhythmic stutter and glitch effects from the frozen spectrum.
When you’d reach for it — You want rhythmic spectral gates, beat-synced stutter effects, chaotic glitch textures, or the ability to sustain a single spectral moment indefinitely at the push of an LFO cycle.
Quick example
- Feed a synth pad into Spectral Freeze Stutter.
- Choose Sqr LFO shape for clean on/off stutters.
- Set Rate to 2 Hz and Threshold to 50% for half-time freezing.
- Reduce Crossfade to 2 ms for sharp glitch cuts, or increase to 20 ms for smoother transitions.
- Add 30% Jitter to randomize exactly when each freeze triggers for a less mechanical feel.
Parameters
| Parameter | What it controls | Range | Sweet spot hint |
|---|---|---|---|
| LFO Shape | Waveform driving the freeze trigger | Sine / Tri / Saw+ / Saw- / Sqr | Sqr for regular stutters; Sine for gradual |
| Rate | LFO speed | 0.05 – 5 Hz | 0.5 – 2 Hz for rhythmic effects |
| Depth | LFO intensity | 0 – 100% | 100% for full effect |
| Threshold | LFO level that triggers a freeze | 0 – 100% | 50% for even on/off duty cycle |
| Crossfade | Transition time between live and frozen | 0 – 50 ms | 2 – 10 ms for glitch; 15 – 30 ms for smooth |
| Jitter | Random variation on threshold per trigger | 0 – 100% | 10 – 40% to avoid mechanical repetition |
| Mix | Wet/dry balance | 0 – 100% | 50 – 100% |
| Gain | Output level | -24 – +12 dB | 0 dB |
Spectral Stutter
What it does — Rapidly repeats short spectral segments at a configurable rate, with per-repetition decay and optional pitch shift.
When you’d reach for it — You’re building EDM fills, glitch transitions, drum-roll build-ups, or vocal chop effects where each repeat can get quieter and shift pitch for rising or falling energy.
Quick example
- Feed a drum hit into Spectral Stutter.
- Set Rate to 8 Hz and Length to 60 ms for fast 16th-note-style stutters.
- Set Count to 8 repetitions with Decay at 0.15 so each repeat fades.
- Set Pitch/Rep to +1 semitone so each repeat rises in pitch, creating a build-up effect.
Parameters
| Parameter | What it controls | Range | Sweet spot hint |
|---|---|---|---|
| Rate | How often a new stutter is triggered | 0.5 – 50 Hz | 2 – 16 Hz for rhythmic patterns |
| Length | Duration of each repeated segment | 10 – 500 ms | 30 – 100 ms for tight stutters |
| Count | Number of repetitions per trigger | 1 – 32 | 2 – 8 for musical rolls; 16+ for machine-gun effects |
| Decay | Volume reduction per repetition (0 = none, 1 = instant) | 0 – 1 | 0.10 – 0.30 for natural fading |
| Pitch/Rep | Pitch shift per repetition | -12 – +12 st | +0.5 to +2 st for rising builds; negative for drops |
| Crossfade | Smoothing frames between repetitions | 0 – 16 frames | 2 – 6 for clean splices |
| Phase | How phase is handled during pitch shift | Preserve / Blend / Dominant / Complex Avg | Blend for general use |
Spectral Smear
What it does — Blurs magnitude along the time axis using a convolution kernel, creating a motion-blur effect for audio.
When you’d reach for it — You want to soften harsh transients into something dreamier, add a trailing wash to percussive sounds, or create ethereal ambient textures where attacks dissolve into smooth pads.
Quick example
- Feed a drum loop into Spectral Smear.
- Set Blur to 25 frames with Gaussian shape and Forward direction — each hit trails off into a soft wash.
- Enable Transients to preserve the initial attack clarity while smearing only the sustain.
- Set Mix to 60% to keep some of the original punch.
Parameters
| Parameter | What it controls | Range | Sweet spot hint |
|---|---|---|---|
| Blur | Width of the smear kernel in frames | 1 – 100 frames | 10 – 30 for subtle smoothing; 50+ for heavy wash |
| Shape | Kernel weight distribution | Box / Gaussian / Exponential | Gaussian for natural falloff |
| Direction | Which way in time the blur extends | Forward / Backward / Symmetric | Forward for trailing tails; Symmetric for even blur |
| Transients | Preserve attack clarity during smearing | On / Off | On when you want smeared sustain but sharp attacks |
| Threshold | Energy jump that counts as a transient (when Transients is on) | -60 – 0 dB | -20 dB |
| Mix | Wet/dry balance | 0 – 100% | 40 – 70% |
| Gain | Output level | -24 – +12 dB | 0 dB |