Patch Changes
-
e6ad386: feat(assemblyai): add the
assemblyai-stt-3-7transcription model and theassemblyai-tts-1speech model- Add
assemblyai-stt-3-7, AssemblyAI's latest pre-recorded transcription model, with prompting, keyterms, temperature, andremoveAudioTagssupport. - Emit a deprecation warning for
universal-3-5-pro, which AssemblyAI has deprecated, and point theuniversal-2notice toassemblyai-stt-3-7. - Stop listing
universal-3-pro, which AssemblyAI has deprecated and no longer includes in its accepted models. - Remove the
bestmodel and the singularspeech_modelrequest parameter it used, which AssemblyAI now rejects.bestandslam-1are sent throughspeech_modelslike any other ID, so AssemblyAI's error lists the accepted models. - Add
assemblyai.speech()for the AssemblyAI text-to-speech API (assemblyai-tts-1), withvoice,language, andoutputFormatmapping,encoding/container/sampleRate/bitRateprovider options, and billed characters reported as speech usage and inproviderMetadata.
- Add