OpenAI TTS is the text-to-speech feature of the OpenAI Audio API, exposed through a speech endpoint that converts written text into lifelike spoken audio. It comes with 11 built-in voices, and a single request takes three inputs: the model, the text to speak and the voice. An optional instructions field lets developers steer delivery, for example asking for a cheerful and positive tone. Output can be saved as an MP3 file or streamed for realtime audio.
It is aimed at developers and product teams who want speech inside their own software rather than a consumer voiceover app. Backend engineers, app builders and prototyping teams can call it with a few lines of code, and official examples are shown in JavaScript, Python, Go and Java. Content teams with an engineering partner can use it to produce audio versions of their writing, and companies building voice features can combine it with the broader audio and realtime tooling in the same documentation.
Typical uses listed include narrating a written blog post, producing spoken audio in multiple languages, and giving realtime audio output using streaming. Developers might add read-aloud buttons to articles, generate voice prompts for an app, or build the speech side of a conversational assistant. Usage policies require a clear disclosure to end users that the voice they hear is AI-generated and not human, which matters for any public-facing product. The same documentation also covers custom voices, voice agents and speech-to-text, so teams can extend a first prototype into a fuller voice experience.
