Radio Now

AI-powered audio broadcast generator with voice cloning.

Features

• Create custom voices from short reference audio clips.
• Generate spoken broadcasts from a topic plus optional creative "twist".
• Configure target script length from the admin panel.
• Optionally generate TTS in paragraph-aligned chunks and stitch with silence.
• Auto-generate and store teaser summaries per broadcast.
• Public listing page with inline audio playback and script view.

Tech Stack

• FastAPI + Jinja2 templates
• MongoDB
• Google Gemini (script + summary generation)
• Replicate qwen3-tts (voice cloning + speech synthesis)
• Docker Compose for local/prod-style deployment

Quick Start

1) Configure environment

Create a local .env file (or export vars in your shell):

RADIONOW_GEMINI_API_KEY=your_gemini_api_key
RADIONOW_REPLICATE_API_TOKEN=your_replicate_api_token
RADIONOW_PORT=8090
# Optional: if set, radionow uses this external/shared MongoDB and does NOT
# start the local mongodb service.
# RADIONOW_MONGO_URL=mongodb://user:pass@host:27017/

# Optional local Mongo credentials (used when RADIONOW_MONGO_URL is unset):
# RADIONOW_MONGO_INITDB_ROOT_USERNAME=admin
# RADIONOW_MONGO_INITDB_ROOT_PASSWORD=admin

# Optional docker network override (defaults to breathnet):
# DOCKER_NETWORK_NAME=breathnet

2) Build and run

make run

Behavior:

• If RADIONOW_MONGO_URL is unset, make run starts a local MongoDB container (localdb profile).
• If RADIONOW_MONGO_URL is set, make run starts only the app and uses that external MongoDB URL.

3) Open the app

• Public page: http://localhost:8090/
• Admin panel: http://localhost:8090/admin

Makefile Commands

These match the repository Makefile:

• make build - Build the Docker image
• make init - Ensure audio/ and voices/ directories exist
• make run - Start services
• make logs - Show application logs
• make shell - Open a shell in the app container
• make stop - Stop services
• make help - Show command help

Project Structure

radionow/
├── app/
│   ├── main.py
│   ├── generator.py
│   ├── models.py
│   ├── database.py
│   ├── templates/
│   └── static/
├── audio/               # Generated broadcast audio (volume)
├── voices/              # Uploaded voice reference audio (volume)
├── docker-compose.yml
├── Dockerfile
├── Makefile
└── requirements.txt

Notes

• This repo intentionally does not contain runtime secrets.
• .env files are gitignored.
• If deploying publicly, place /admin behind authentication and HTTPS.
• The stack uses an external Docker network (breathnet by default). make init/make run creates it automatically when missing.