Student-driven radio engineering, local AI audio tools, and resilient broadcast systems from TED University.
RadioTEDU is TED University's student radio and media laboratory. RTAI is its engineering family for local-first broadcast automation, listener interaction, mobile experiences, and responsible generative-audio tooling.
This account contains the maintained public engineering portfolio. Each repository documents its own requirements, operational boundaries, and verification workflow.
| Project | Purpose |
|---|---|
| RTAI Radio | Bilingual AI radio, durable station orchestration, listener publishing, metadata, and Windows broadcast operations |
| RTAI Mobile | React Native listener app, Android Auto integration, Study, voting, and Jukebox controller experiences |
| VoterTAI | Listener-controlled next-song voting with an authoritative backend and supervised local playout agent |
| RTAI Jingle | Privacy-first Windows studio for local QwenTTS narration and automatic music mixing |
| RadioTEDU profile | Account branding, portfolio navigation, and cross-project context |
RTAI Jingle ── creates lawful narration and station imaging
│
▼
RTAI Radio ─── programs and publishes the bilingual radio service
│
├──────────────► RTAI Mobile ── listener and Study experiences
│
└──────────────► VoterTAI ───── next-song voting and local playout
The repositories remain independently deployable. Product names describe their roles; internal package names, application identifiers, station identities, and service names remain stable unless a project documents a migration.
- Local-first intelligence: models and media processing stay local where the project architecture supports it.
- Operational clarity: runbooks, health checks, fallbacks, and explicit handoffs are part of the product.
- Privacy and consent: repositories document model downloads, secrets, telemetry behavior, and media-rights boundaries.
- Evidence over demos: public documentation distinguishes implemented behavior from planned work.
- Student ownership: the portfolio is built as a practical radio, broadcasting, and software-engineering laboratory.
The portfolio uses Python, TypeScript, React, React Native, Vite, Liquidsoap, Icecast, SQLite/PostgreSQL, local Ollama models, QwenTTS, FFmpeg, and Windows service tooling. Each repository's README is the authority for its supported runtime and installation process.