Real-Time Weather & Audio Reactiveness
Overview
This project combines local weather data and audio input into an ambient visualisation system that mirrors emotional tone through visuals. It maps real-time weather conditions alongside audio features to produce an environment that responds like a living system.
The system works across two parallel paths that merge into a single visual output:
Audio In → Feature Extraction → Emotional Mapping ──┐
↓
Mood Merge → Visual Logic
↑
Weather API → Environmental Normalisation ────────────┘
Audio Analysis
The audio path extracts:
- Rhythm — tempo and beat structure
- Pitch — fundamental frequency and harmonic content
- MFCCs — Mel-frequency cepstral coefficients for timbre
- Intensity — RMS amplitude
- Spectral contrast — brightness and darkness in the spectrum
- Chroma — pitch class distribution
These features feed an emotional classification model that assigns mood weights — calm, tense, euphoric, melancholy — in real time.
Weather Analysis
Weather data was pulled via Python API and normalised:
- Temperature — scaled to a warmth/coolness visual axis
- Humidity — affects visual density and diffusion
- Wind speed and angle — influences directional motion
Merged Output
Both paths converge in TouchDesigner. The combined mood state determines colour palette, particle behaviour, motion speed, and visual density simultaneously. The output feels like a space that knows what’s happening — inside and outside.
Learning Outcomes
Merging two seemingly unrelated data sources — weather and sound — into a coherent aesthetic system required building a shared emotional vocabulary. The key insight: both weather and music carry mood. The system treats them as two instruments in the same piece.