OpenAI.fm is OpenAI's own interactive demo for its text-to-speech API. Type or paste any text, pick from a set of built-in voices, and steer the delivery with a written style prompt — 'calm and reassuring', 'excited sports announcer', that kind of direction — then play the result back or download the audio file.

It's built for developers deciding whether to integrate the Speech API into their own product, so the interface stays close to the API's actual parameters: voice selection, instructions text, and format output map directly onto what you'd send in a real request. The source code is open on GitHub, so teams can see exactly how the demo calls the API rather than treating it as a black box.

The demo itself is free to use with no account required. Building the same functionality into your own app means paying for the underlying OpenAI Speech API on a per-character or per-minute basis, billed like any other OpenAI API usage. There's no separate consumer subscription for OpenAI.fm — it exists purely to let you hear the voices before you commit to integrating them.