openai-edge-tts

Why we included this project

Teams already paying for OpenAI's speech endpoint can switch their base URL to this local service and keep their existing request code mostly untouched. It mimics the /v1/audio/speech route, so the same voice names and speed options work, while the audio itself comes from Microsoft Edge's neural voices, which are free and have no per-character pricing. The standard OpenAI voice names map to Edge equivalents, and you can also pass any edge-tts voice directly if you need more variety. SSE streaming and output formats like mp3, opus, aac, flac, wav, and pcm are supported, and the official Docker image gets a self-hosted instance running quickly, which suits both prototyping and production use.