Groq Orpheus text to speech
- Generate English and Saudi Arabic speech through your existing Groq key and the normal audio router.
- Return validated PCM16 WAV, normalize Groq's streamed WAV headers, and retain existing failover and rate-limit backoff.
- Standard OpenAI voice names map to tested native defaults; explicit native voices are supported.
- Add regression coverage for streamed audio, malformed responses, model terms, failover and signed-catalog audio imports.
Models arrive through the signed catalog. Premium receives the live catalog; Free receives new rows after the unchanged 30-day aging period. Your Groq organization administrator must accept each Orpheus model's terms. TwelveLabs and Mistral are unchanged in this release.
Includes the already-merged Anthropic-compatible Messages endpoint improvements from #1821, with contributor credit preserved in the README.