Tacitum

Tacitum is a real-time browser audio visualizer for conversational listening states.

It transforms live audio into two expressive modes:

• DOT: a ripple-driven depth plane with controllable lift and flow behavior
• JOY: a topographic field for broader ambient shape and energy

Built with TypeScript + Vite, Tacitum is designed for fast local iteration, deterministic model tests, and production deployment on Vercel. Its audio, speech, and visual layers are decoupled so they can be lifted into future tools that turn sound input into visualization.

Live Demo

• Production demo: https://tacitum.vercel.app

Screenshots

DOT mode — depth plane, ripple-driven

JOY mode — topographic field

Core Features

• Real-time microphone or shared tab-audio analysis and visualization
• Two visual modes (DOT / JOY)
• Fine-grained controls for sensitivity, lift, speed, zoom, and browser audio source
• Camera controls with drag rotation, zoom, and optional ambient auto motion
• Persistent control state in local storage with migration safeguards
• Strong unit test coverage for audio math, visual geometry, camera behavior, and state machines

Tech Stack

• TypeScript
• Vite
• Vitest

Getting Started

Prerequisites

• Node.js 20+
• npm 10+

Install

npm install

Run local development

npm run dev

Run tests

npm test

Build for production

npm run build

Project Structure

• src/app: app shell and control panel wiring
• src/audio: input, analysis, and signal math
• src/speech: speech-state modeling and buffering
• src/visual: renderer, camera, geometry, and DOT/JOY field models
• src/types: shared domain types
• docs: implementation specs and screenshots

Reusable Parts

Tacitum is structured so its core layers can be reused independently in other sound-driven projects:

• src/audio: microphone and tab-audio capture, FFT analysis, and sensitivity math — a self-contained signal pipeline.
• src/speech: speech-state modeling and event buffering that turns raw audio into higher-level listening states.
• src/visual: renderer, camera, geometry, and the DOT/JOY field models that map signal energy to motion.

Each layer has deterministic unit tests, so you can adopt one without pulling in the full app shell.

Status

Tacitum is production-deployed and actively refined for visual continuity, smooth camera behavior, and externally shareable repository quality.