No.148 · AI & Chat
Voice Orb
A voice conversation reduced to one object: a soft sphere whose clouds breathe with your voice, gather while it thinks and pulse as it speaks — with one line of caption underneath.
The prompt
The design brief behind this component — reuse it to regenerate, remix, or explain it.
Design a minimal, monochrome voice-mode screen: a white (or near-black) field, one sphere, one caption line, two round controls. The sphere (clamp 10–13rem) has a pale blue radial base, three large blurred 'clouds' (blur 18px, multiply blend on light) drifting on 7s / 9s / 11s loops, and a soft top-left sheen. The conversation state changes behaviour, not icons. A single level signal (0–1) scales the sphere up to 7%: while listening it follows the user's voice (synthetic here, an AnalyserNode in production); while thinking the sphere contracts to 0.94 and the clouds turn much faster, as if gathering; while speaking the level follows the reply's syllables and the clouds speed up a little; idle is a slow 5s breath. Muting desaturates the sphere. The caption says 'Listening…', then shows the transcribed question, then streams the reply word by word. Controls: a mic mute toggle (inverts when muted) and a red end button.
Notes
- Interaction
- Tap the sphere to start; mute or end with the controls below.
- Motion
- Per-frame level (lerped) drives scale; cloud drift loops change speed by phase; caption lines rise in over 420ms.
- Accessibility
- The sphere is a button with a state-aware label; captions are a polite live region; mute is a toggle with aria-pressed. Reduced motion stills the clouds and level.
- Responsive
- The sphere scales with the viewport; controls stay 52px.
- On touch
- Designed for taps; nothing depends on hover.