Together AI introduces pronunciation_dict parameter for TTS

AI Impact Summary

The Together AI platform is introducing a new `pronunciation_dict` parameter for TTS requests, allowing users to override model pronunciations for specific tokens. This enables greater control over the generated speech, particularly useful for handling domain-specific terminology or unusual words. The introduction of this parameter highlights a shift towards greater customization and flexibility in the TTS service, though it doesn't directly address any deprecations or model changes.