valence · arousal · dominance
An ambient generator built the way Brian Eno built Music for Airports: a handful of Web Audio voices, each looping on its own non-matching cycle length, drifting in and out of phase forever. What's new here is the control surface — drag the dot inside the triangle to set a mood in Valence/Arousal/ Dominance space, and the pitch, tempo, register and brightness of every voice bend toward it live, while it plays.
the ensemble
| voice | type | pitch | frequency | loop |
|---|
base tone (binaural)
headphones on: this plays two plain sine tones, one per ear, a few Hz apart. Your brain perceives the difference as a slow pulsing beat at that rate — an actual binaural-beat illusion, layered underneath the ensemble, not a claim about what it does to you.
Each corner is a pole: pull toward VALENCE for a brighter, more consonant drone; toward AROUSAL for a faster, denser pulse; toward DOMINANCE for a deeper, more forward, drier sound. The centroid is neutral.
Under the hood: distance from the valence corner sets tension — every voice's note glides between a consonant pentatonic degree and a dissonant one (a minor 2nd, a tritone), so an unhappy drag bends the pitch there instead of jump-cutting. Distance from arousal sets the shared base tempo that each voice's irrational cycle-length multiplier is built on. Distance from dominance sets the root register (how deep the bass is) and how dry vs. reverberant the mix sits.
The ensemble is two pads, a bell, a chime, two wordless vox voices (sawtooths carved into different vowel-like formant shapes) and a low hum, each on its own irrational cycle length so they drift in and out of phase forever — the table's loop column shows how long each voice's current cycle is, live. The optional base tone underneath is a separate binaural beat: unrelated to the VAD mood, just a steady brainwave-band pulse you can layer in or leave off.
No samples, nothing server-side — it's oscillators, filters and a convolution reverb, synthesized live in your browser.