cross-posted from: https://reddthat.com/post/71512059
I’m planning to self-host a 24/7 radio show and need some architecture advice.
The vision:
- Monitor my RSS feeds and social media posts, then turn them into spoken-word segments.
- Follow a fixed daily schedule (e.g., news at 8 AM, sports/tech at 9 AM, music blocks, etc.).
- Serve the stream via a web player on my site.
- Keep audio files on disk for three days so I can catch up on anything I missed after airing, with the option to choose playback speed.
The confusion:
Are any of the tools I’m considering actually suitable for audio streaming and playlist management? Or am I approaching this the wrong way?Current thinking:
- Use
radio-djorSUB/WAVEas the broadcast engine.- Use Piper TTS for speech synthesis.
- Use Cron for scheduling.
- Write a custom Python script with Ollama (local) for content generation, in case I want an AI “host” to create a continuous flow from all the disparate sources.
Has anyone built something similar? What framework would you recommend? Am I overlooking a dedicated FOSS radio automation tool that already does this?
Thanks in advance!


I would suggest using llama.cpp instead of Ollama, or maybe Unsloth Studio
https://sleepingrobots.com/dreams/stop-using-ollama/
!localllama@sh.itjust.works