Hourly AI Radio Newscaster
Budget: $10 – $55 USD
I run a small-scale internet radio station and want an AI voice that automatically joins the stream for a three-minute world-news break at the top of each hour. The core requirements are straightforward:
• Content source – pull the latest headlines and short summaries on-the-fly from Worthy News and Rapture Ready News (RSS or API is fine). If one feed is down the system may fall back on any publicly available source you feel fits the same style.
• Scheduling – simple on/off toggle plus a field where I can list specific hours (e.g., 08:00-22:00) when the bulletin should air; outside those hours it stays silent.
• Custom intro – every bulletin starts with a short, scripted greeting that states the AI reporter’s chosen name and my station name before moving into the stories. I’ll supply the exact wording.
• Runtime – text must be fetched, interpreted and voiced in real time so it never sounds pre-recorded. A natural-sounding TTS engine is essential.
• Output – send the finished audio into my existing stream (Icecast/Shoutcast compatible) without interrupting anything else that’s playing.
• Control – web panel or simple config file where I can update the intro script, change source URLs, and adjust the schedule.
• Budget – this is a lean $55 build, so keep the solution light, preferably using open-source libraries.
The voice model must sound human and tts be setup on my server. i dont want monthly fees we can use open source models.
Delivery is the working code, brief setup guide, and a quick test session to prove the hourly hand-off works.
• Content source – pull the latest headlines and short summaries on-the-fly from Worthy News and Rapture Ready News (RSS or API is fine). If one feed is down the system may fall back on any publicly available source you feel fits the same style.
• Scheduling – simple on/off toggle plus a field where I can list specific hours (e.g., 08:00-22:00) when the bulletin should air; outside those hours it stays silent.
• Custom intro – every bulletin starts with a short, scripted greeting that states the AI reporter’s chosen name and my station name before moving into the stories. I’ll supply the exact wording.
• Runtime – text must be fetched, interpreted and voiced in real time so it never sounds pre-recorded. A natural-sounding TTS engine is essential.
• Output – send the finished audio into my existing stream (Icecast/Shoutcast compatible) without interrupting anything else that’s playing.
• Control – web panel or simple config file where I can update the intro script, change source URLs, and adjust the schedule.
• Budget – this is a lean $55 build, so keep the solution light, preferably using open-source libraries.
The voice model must sound human and tts be setup on my server. i dont want monthly fees we can use open source models.
Delivery is the working code, brief setup guide, and a quick test session to prove the hourly hand-off works.