AssemblyAI Dictation as a cloud transcription option

[dan-ince-aai]

AssemblyAI's Dictation API as a cloud transcription option. It returns the transcript already cleaned up, so you get text that's ready to send without a separate AI enhancement call.

• One request: POST https://dictation.assemblyai.com/v1/transcribe with the recording as a multipart audio part (WAV, raw PCM, MP3, M4A, Ogg/Opus, FLAC, WebM; 80 ms to 20 minutes) and an Authorization: <api key> header. Raw PCM needs sample_rate and channels in the config.
• The response has text (verbatim) and llm_response (filler words removed, punctuation, capitalization and self-corrections fixed). Use llm_response, falling back to text when it's empty.
• An optional JSON config part takes keyterms_prompt (names and jargon to spell right) and language_codes. It covers 32 languages.
• It isn't OpenAI-compatible, so it needs its own small request rather than the OpenAI-compatible path.

Looking for a maintainer's view on whether this fits before I write anything. I'm happy to do the issue and PR once it's approved.

Disclosure: I work at AssemblyAI. If you have questions or want a hand with anything on our side, comment here or reach me on X at @dan_aai.