Support OpenAI verbose_json for /v1/audio/transcriptions (+ multipart field fix) - #510

Open

ademasi wants to merge 5 commits into

ademasi:upstream/whisper-verbose-json

Commits on Jul 5, 2026

fix(server): forward multipart audio fields to transcription handler

ademasi
and
claude
committed
71abe2b
feat(server): support OpenAI verbose_json for audio transcriptions

ademasi
and
claude
committed
f4672a8
fix(server): defensive temperature parse in multipart dispatch

ademasi
and
claude
committed
e1c2516
refactor(server): hoist raw_output/language above transcription #ifdef

ademasi
and
claude
committed
90c6a51
fix(server): fallback segment for verbose_json with <2 markers

ademasi
and
claude
committed
17ad979