Podcaster generates a fresh audio briefing on any subject — on demand or on a schedule — running on your own server, with your keys and your data. Give listeners accounts, subscriptions and offline playback. Nothing leaves your machine unless you share it.
A real listener-facing platform you host yourself — not a service you rent. Your subjects, your audience and your audio stay with you.
Runs entirely on your own machine or in a container. No SaaS, no account on someone else's platform, no middleman between you and your audio.
Subjects, listeners and listening history never leave your server. An episode becomes shareable only when you mint a private link for it.
Describe a topic — tech, games, geopolitics, science, finance, sport — and get a fresh, sourced news episode in minutes.
Grounded web search pulls the latest developments, with RSS and a plain-LLM fallback. It remembers what it covered so episodes don't repeat.
Bring any one LLM key; providers fall back automatically and audio is free via Edge TTS. Point it at a local Ollama model to go fully offline.
Run subjects daily or weekly. A single hourly cron entry — or the bundled scheduler container — fires whatever is due.
Install it like an app, tap Save offline, and your episodes play with no connection — on the train, on a plane, anywhere.
Run it just for yourself or open it to a household or team — admin, publisher and reader roles, with publishing gated.
Readers subscribe to the subjects they care about and get a private listen link by email whenever a new episode drops.
French, English, Spanish, German, Italian — native voices, with one-click regeneration of any episode into another language.
The real app — captured straight from the running server.
Everything — manage users & roles, API keys, the SMTP setup and the scheduler. The first registered account.
Create subjects, generate episodes on demand and schedule them. Cannot touch users or system settings.
The default. Listen, save episodes offline, subscribe to subjects and get notified by email. No publishing.
Ship it in a container, or run it straight from Python. You only need one LLM key to start — audio works for free via Edge TTS.
# 1 — configure your keys cp .env.example .env # add at least GEMINI_API_KEY in .env # 2 — build & run docker compose up -d # or: podman-compose up -d # 3 — open the app http://127.0.0.1:8077
scheduler service fires due schedules hourly — no host cron needed.python -m venv .venv && source .venv/bin/activate pip install -r requirements.txt cp .env.example .env # add a key python server.py # http://127.0.0.1:8077 # generate from the CLI too: python generate.py markets-finance --lang en --minutes 12
Grounded web search gathers the latest developments for the subject, skipping anything already covered.
An LLM turns the brief into a clear, structured episode at your chosen length and language.
It's voiced with native per-language voices and saved as an MP3 you own.
It lands in the catalog, retention prunes old episodes, and subscribers get an email.