POST
https://your-instance.com/v1/audio/speechReturns raw audio bytes rather than JSON. Write the body straight to a file, or stream it to the client.
modelstringrequiredSpeech model to use.
inputstringrequiredThe text to speak.
voicestringrequiredWhich voice to use. Available voices vary by model.
response_formatstringdefault: "mp3"Audio container: mp3, opus, aac, flac, wav or pcm.
speednumberdefault: 1.0Playback rate between 0.25 and 4.0.
curl https://your-instance.com/v1/audio/speech \
-H "Authorization: Bearer $CLAWROUTER_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "tts-1",
"input": "The quick brown fox.",
"voice": "alloy"
}' \
--output speech.mp3