Radio Starlight
Radio Starlight is a generative radio tool that reinvents the traditional radio experience. It offers personalized voice talks about songs ...
Last verified:
What is Radio Starlight?
Radio Starlight is a generative radio-station builder that turns web content, APIs, feeds, weather and music into on-demand, narrated radio programs using configurable segments and AI models. Users design a program once by laying out segments, binding each segment to one or more sources (RSS/Atom/JSON, web pages, APIs, weather, or music), and attaching per-segment prompts and background music to produce consistent episodes. The platform supports both cloud LLMs (for example via user-provided OpenAI API keys) and on-device Apple Foundation Models for text and image generation, plus an image playground to create cover art and avatars. Designed for creators, podcasters, hobbyist broadcasters, and developers, it also exposes an open publishing protocol (StarlightCatalogService) and exportable JSON so shows, personalities, and sound sets can be inspected, remixed, and shared.
Radio Starlight pricing
Pricing model: Free
The website describes a Basic free tier with core app functionality and a Pro level that unlocks advanced Custom features; many generation features require the user's own API key (so API usage costs are separate), and on-device model usage depends on your Apple device and OS. Specific paid-plan names, monthly fees, or tiered pricing amounts are not listed on the site; instead, the site states that Pro unlocks custom behaviors and that users supply their own model API credentials when using cloud LLMs.
Radio Starlight pros
- Segment-first design for repeatable show structure
- Supports multiple source types: RSS, Atom, JSON feeds
- Can scrape and summarize arbitrary web pages per segment
- API content binding to include dynamic external data
- Live weather integration for localized forecasts in shows
- Music scheduling with support for playlists and licensed libraries
- Per-segment prompts allow fine-grained voice and content control
- Works with cloud LLMs (user-provided API keys) for flexible model choice
- On-device Apple Foundation Models support offline or private generation
- Image generation for cover art and personality avatars
- Open StarlightCatalogService protocol for publishable, inspectable shows
- Import/export of Programs, Personalities, and Sound Sets as JSON
- Background music can be swapped between segments for variety
- Generates multilingual summaries and narration from various sources
- Reference implementation and specification available as open source on GitHub
Radio Starlight cons
- Requires user-provided OpenAI/API keys for some cloud model features
- On-device generation availability depends on Apple hardware and OS
- No bundled model credits — external API costs are the user's responsibility
- Music licensing and integration details are not fully automated
- Live scraping of pages may produce inconsistent source formatting
- Feature availability differs by platform (macOS vs iOS) and version
- Exported programs rely on external sources remaining available for full replay
- Learning curve to design effective per-segment prompts and formats
Frequently asked questions about Radio Starlight
How do I add news and web content into a program?
You add sources per segment by pointing the segment to RSS/Atom/JSON feeds, calling APIs, or providing web page URLs; Radio Starlight will fetch, summarize, and incorporate that content into the segment using the segment-specific prompt you supply.
What AI models can Radio Starlight use to generate speech and text?
Radio Starlight supports cloud LLMs via user-provided API keys (for example OpenAI) and can also use Apple Foundation Models on-device where available, letting you choose cloud or local generation depending on privacy, cost, and hardware.
Can I include live weather in my broadcasts?
Yes — Radio Starlight can pull live weather and forecasts as a distinct source and bind them to segments so programs include up-to-date local conditions and forecasts.
How do I control the on-air personality and voice?
Design a Personality and attach it to your program; per-segment prompts and personality settings control style, tone, and how the AI discusses incoming content, and you can also create visual avatars and cover art for that personality.
Is there a way to share or publish the shows I create?
Yes — Radio Starlight exposes an open publishing protocol (StarlightCatalogService) and lets you export designed programs, generated episodes, personalities, sound sets, and the exact sources used so others can inspect, remix, or re-run your programs.
Do I need to pay for using Radio Starlight’s AI generation?
The app itself offers a Basic free tier, but cloud model usage requires your own API key and will incur charges from the model provider; the site indicates a Pro tier unlocks custom features but does not list bundled generation credits.
Can I run Radio Starlight on mobile devices?
Mobile access is available through the companion Starlight GO player for iOS, which retrieves programs (for example via iCloud) created on macOS, but full creation and some advanced features are best experienced on macOS with the main app.
What file formats can I export or import?
Programs, Personalities, and Sound Sets can be imported and exported as JSON files so you can share, version, and programmatically manipulate station definitions and assets.
How does music integration work?
You can schedule music playlists or connect licensed music libraries as segment sources; music is treated as a source per segment and you can swap background tracks between segments for different moods, though licensing and playback specifics depend on the connected library.
Is Radio Starlight open source or does it provide developer docs?
Yes — the site links to an open specification and a reference implementation on GitHub, and documents the StarlightCatalogService schema so developers can build interoperable tools and services around exported catalogs.