k2k audio Blend two sounds into one. Like double-exposing two photographs — morph timbres, swap spectral shapes, crossfade by frequency, or transfer the character of one sound onto another.
Visual Blend
What it does — Combines two spectral inputs using classic image-style blend modes: additive, multiply, maximum, minimum, screen, or difference.
When you’d reach for it — You have two processed versions of a sound (say, extracted transients and a smoothed sustain) and need to recombine them, or you want to layer two spectra with precise control over how overlapping energy interacts.
Quick example
- Connect your two sources to Input A and Input B.
- Set the blend mode to Additive for a natural layering.
- Dial Mix to 0.50 for an even balance.
- Switch Phase to Dominant so the louder source always drives phase.
- Enable Loop B if the second source is shorter and you want it to repeat.
Parameters
| Parameter | What it controls | Range | Sweet spot hint |
|---|---|---|---|
| Mix | Balance between A and the blended result | 0.00 — 1.00 | 0.50 for equal weight |
| Blend mode | How magnitudes combine (Additive, Multiply, Maximum, Minimum, Screen, Difference) | 6 modes | Additive for layering, Multiply for masking |
| Phase | Which source provides phase information (From A, From B, Interp, Dominant) | 4 modes | Dominant keeps the cleaner phase |
| Loop B | Whether B repeats when shorter than A | on / off | On when B is a short texture |
Visual Morph
What it does — Fits B-spline curves to the spectral peaks of both sources, then reshapes A’s peak profile toward B’s using ratio correction and phase reconstruction.
When you’d reach for it — You want A to gradually take on B’s spectral shape without a simple crossfade — the peaks themselves migrate, keeping a sense of natural transition rather than volume blending.
Quick example
- Feed a vocal into Input A and a cello into Input B.
- Set Morph to 0.30 for a subtle reshaping.
- Listen — A’s formants shift toward the cello’s resonance profile.
- Push Morph higher (0.70+) for a more dramatic transformation.
Parameters
| Parameter | What it controls | Range | Sweet spot hint |
|---|---|---|---|
| Morph | How far A’s spectral profile moves toward B’s | 0.00 — 1.00 | 0.30 — 0.50 for recognizable hybrids |
| Loop B | Whether B repeats when shorter than A | on / off | On for short reference textures |
Spectral Morph
What it does — Decomposes both sources into harmonic, percussive, and residual layers (HPSS), then lets you blend each layer independently between the two inputs.
When you’d reach for it — You want drum attacks from one source with the tonal sustain of another, or you need to swap just the noise floor between two recordings while keeping everything else intact.
Quick example
- Connect a drum loop to Source A and a synth pad to Source B.
- Set Percussive to 0.00 (keep A’s attacks) and Harmonic to 1.00 (take B’s tonal body).
- Leave Residual at 0.50 for a blended noise floor.
- Raise Separation if the layers bleed into each other too much.
Parameters
| Parameter | What it controls | Range | Sweet spot hint |
|---|---|---|---|
| Harmonic | Blend position for tonal/harmonic content | 0.00 — 1.00 | 0 = A’s harmonics, 1 = B’s harmonics |
| Percussive | Blend position for transient/percussive content | 0.00 — 1.00 | 0 = A’s attacks, 1 = B’s attacks |
| Residual | Blend position for noise/residual content | 0.00 — 1.00 | 0.50 for natural blending |
| Separation | How aggressively HPSS separates the layers | 0.00 — 1.00 | 0.70 — 1.00 for clean separation |
| B loop | Whether B repeats when shorter than A | on / off | On when sources differ in length |
Timbre Morph
What it does — Reshapes A’s timbral character toward B by computing perceptual differences (centroid, spread, tilt, flatness) and applying scaled corrections to A’s own spectrum.
When you’d reach for it — You want a flute to gradually sound more like an oboe without replacing it — the pitch, rhythm, and structure stay from A, but the “color” shifts toward B.
Quick example
- Connect the flute to Input A and the oboe to Input B.
- Set Morph to 0.40 for a partial timbral shift.
- Raise Harmonics weight to emphasize the harmonic recipe change.
- Lower Noise weight if the noise floor transformation is too aggressive.
- Adjust Gain if the output level has drifted.
Parameters
| Parameter | What it controls | Range | Sweet spot hint |
|---|---|---|---|
| Morph | Master intensity of all timbral corrections | 0.00 — 1.00 | 0.30 — 0.60 for musical results |
| Harmonics | Weight of the harmonic ratio correction | 0.00 — 1.00 | 1.00 to fully reshape harmonic recipe |
| H/P/R | Weight of harmonic/percussive/residual rebalancing | 0.00 — 1.00 | 1.00 for full component rebalancing |
| Decay | Weight of temporal decay rate morphing | 0.00 — 1.00 | 0.50 — subtle changes to sustain character |
| Noise | Weight of noise floor spectral shaping | 0.00 — 1.00 | 0.50 to avoid over-shaping quiet regions |
| Gain | Output level adjustment | -24 dB — +24 dB | 0 dB unless compensating for level drift |
| Phase | How phases are combined (Preserve, Blend, Dominant) | 3 modes | Dominant for cleanest results |
| Loop B | Whether B repeats when shorter than A | on / off | On when sources differ in length |
Spectral Vocoder
What it does — Imposes the modulator’s per-band amplitude dynamics onto the carrier’s spectrum, creating classic vocoder effects where one sound “speaks” with another’s voice.
When you’d reach for it — You have a harmonically rich carrier (synth pad, strings) and want a voice or drum pattern to shape its dynamics in real time — talking synths, rhythmic filtering, robotic vocals.
Quick example
- Connect a synth pad to the Carrier input and a vocal to the Modulator input.
- Set Depth to 1.00 for full vocoding.
- Adjust Formant to shift the vocal character up or down in semitones.
- Enable Swap if you accidentally connected the inputs backward.
Parameters
| Parameter | What it controls | Range | Sweet spot hint |
|---|---|---|---|
| Depth | How strongly the modulator shapes the carrier | 0.00 — 1.00 | 1.00 for classic vocoder; lower for subtle rhythm |
| Formant | Shifts the modulator’s spectral envelope up/down | -12 — +12 semitones | 0 for natural; +2 to +4 for chipmunk effect |
| Gain | Output level | 0.00 — 2.00 | 1.00 unless compensating |
| Swap | Exchanges carrier and modulator inputs | on / off | Off by default |
| Mod loop | Whether the modulator loops when shorter than carrier | on / off | On when modulator is a short phrase |
Spectral Convolution
What it does — Captures the average or peak spectral shape of an impulse sound, then stamps that shape onto a main signal by multiplying their spectra bin-by-bin.
When you’d reach for it — You want to color one sound with the spectral fingerprint of another — make a piano sound like it was recorded through a guitar body, or imprint a room’s resonance profile onto a dry recording.
Quick example
- Connect your dry signal to Main and the reference sound to Impulse.
- Choose Average capture for a smooth overall shape, or Peak for a brighter, more aggressive stamp.
- Set Mix to 1.00 for full imprinting.
- If the output is too quiet, raise Normalize toward 1.00 to restore the original peak level.
Parameters
| Parameter | What it controls | Range | Sweet spot hint |
|---|---|---|---|
| Capture mode | How the impulse envelope is measured (Average or Peak) | 2 modes | Average for smooth; Peak for bright/aggressive |
| Imp Gain | Pre-gain on the impulse envelope before application | -24 dB — +24 dB | 0 dB for neutral |
| Normalize | How much to restore the original peak level after imprinting | 0.00 — 1.00 | 1.00 to match input loudness |
| Mix | Dry/wet balance | 0.00 — 1.00 | 1.00 for full effect |
| Out Gain | Final output level adjustment | -24 dB — +24 dB | 0 dB unless compensating |
Formant Morph
What it does — Interpolates the vocal formant structure between two sounds using cepstral envelope extraction, creating smooth vowel-like transitions in log domain.
When you’d reach for it — You want one vocal to gradually take on the vowel shapes of another, or you need to morph the resonant body of one instrument into another while preserving the original excitation (pitch and harmonics).
Quick example
- Connect an “ah” vowel to Input A and an “ee” vowel to Input B.
- Set Morph to 0.50 for a halfway vowel.
- Keep Excitation on Source A so the pitch stays anchored.
- Raise Order if the formant shape needs more detail (higher = sharper formants).
Parameters
| Parameter | What it controls | Range | Sweet spot hint |
|---|---|---|---|
| Morph | How far the formant envelope moves from A toward B | 0.00 — 1.00 | 0.50 for a balanced hybrid vowel |
| Order | Cepstral envelope resolution (lower = broader formants, higher = sharper) | 2 — 40 | 20 for typical vocals; 8 — 12 for broad instrumental shapes |
| Excitation | Which source provides the pitch/harmonic structure (Source A, Source B, Blend) | 3 modes | Source A to keep A’s pitch |
| Phase | How phases are combined (Preserve, Dominant, Blend) | 3 modes | Dominant for cleanest results |
LFO Crossfade
What it does — Crossfades between two audio sources using an LFO, creating rhythmic pulsing transitions with selectable waveform shapes and crossfade curves.
When you’d reach for it — You want an automated, rhythmic back-and-forth between two sounds — tremolo-style gating, DJ-style crossfades, or a breathing pulsation between textures.
Quick example
- Connect two textures to Input A and Input B.
- Choose Triangle for a smooth back-and-forth sweep.
- Set Rate to 0.50 Hz for a slow breathing effect.
- Set Bias to 0.50 so the crossfade centers equally between both sources.
- Switch Curve to Equal Power so energy stays constant through the transition.
Parameters
| Parameter | What it controls | Range | Sweet spot hint |
|---|---|---|---|
| Shape | LFO waveform (Sine, Triangle, Saw Up, Saw Down, Square, S&H, Noise) | 7 shapes | Sine for smooth; Square for hard chops |
| Rate | LFO speed | 0.01 — 50.00 Hz | 0.25 — 2.00 Hz for musical pulsation |
| Depth | LFO modulation intensity | 0.00 — 1.00 | 1.00 for full swing; lower for gentle undulation |
| Bias | Resting crossfade position (0 = all A, 1 = all B) | 0.00 — 1.00 | 0.50 for equal center |
| Curve | Crossfade gain law (Linear, Equal Power, S-Curve, Log) | 4 curves | Equal Power for constant energy |
Spectral Magnitude Crossfade
What it does — Crossfades between two spectra with an LFO that sweeps across frequency, so different frequency bands transition at different times, creating a wave-like spectral motion.
When you’d reach for it — You want something more alive than a flat crossfade — a sweep that rolls through the spectrum, revealing B’s bass before its treble (or vice versa), adding spatial movement to a blend.
Quick example
- Connect two sources to Input A and Input B.
- Set the sweep direction to Low-High so bass frequencies crossfade first.
- Set Rate to 0.50 Hz and Spread to 2.0 for a wide, slow spectral wave.
- Adjust Center to focus the sweep origin around 1000 Hz.
Parameters
| Parameter | What it controls | Range | Sweet spot hint |
|---|---|---|---|
| Shape | LFO waveform for the sweep (Sine, Triangle, Saw Up, Saw Down, Square) | 5 shapes | Sine for smooth rolling |
| Rate | LFO speed | 0.01 — 50.00 Hz | 0.25 — 1.00 Hz for audible sweeps |
| Depth | How far the LFO pushes the mix from the bias position | 0.00 — 1.00 | 1.00 for full crossfade range |
| Bias | Resting mix position | 0.00 — 1.00 | 0.50 for centered |
| Sweep direction | Which frequencies lead the crossfade (Low-High, High-Low, Center) | 3 modes | Low-High for rising sweep |
| Center | Frequency anchor for the sweep pattern | 50 — 10000 Hz | 1000 Hz for a balanced midpoint |
| Spread | How many octaves the phase offset covers | 0.0 — 4.0 | 2.0 for a wide, visible sweep |
| B loop | Whether B repeats when shorter than A | on / off | On when sources differ in length |
Multi-Band Crossfade
What it does — Splits the spectrum into 2—8 bands and crossfades each band independently, driven by spectral similarity between the two inputs — similar regions blend freely, dissimilar regions lock toward your bias preference.
When you’d reach for it — You want a content-aware blend where the spectral material itself decides how to merge. Regions that can morph naturally do; regions that would clash stay clean. Great for hybrid sound design where brute-force blending creates artifacts.
Quick example
- Connect two sources to Input A and Input B.
- Set Bands to 4 for a bass / low-mid / high-mid / treble split.
- Drag the bias points in the curve widget: pull bass toward A, push treble toward B.
- Set Similarity to 0.70 so the content steers the mix but your bias still anchors it.
- Choose Equal Power curve for constant energy.
Parameters
| Parameter | What it controls | Range | Sweet spot hint |
|---|---|---|---|
| Bands | Number of frequency bands | 2 — 8 | 3 — 4 for broad control; 6+ for surgical |
| Similarity | How much spectral similarity steers the crossfade vs. your static bias | 0.00 — 1.00 | 0.70 for content-aware with user anchoring |
| Bias curve | Per-band A/B preference (interactive widget, one point per band) | 0 = A, 1 = B per band | Drag to taste |
| Slope | Crossover steepness between bands | 6 / 12 / 24 / 48 dB | 24 dB for clean separation |
| Curve | Crossfade gain law (Linear, Equal Power, S-Curve, Log) | 4 curves | Equal Power for constant energy |
| Phase | Phase source (Dominant, Source A, Source B, Interpolate) | 4 modes | Dominant for cleanest results |
| B loop | Whether B loops or plays once | on / off | On when sources differ in length |