Requires run. Turns a short voice recording into text, as the app's dictation does. Send the audio itself as the request body with its type in Content-Type (audio/wav, audio/webm, audio/ogg, audio/mpeg, audio/mp4, audio/aac or audio/flac), at most 10 MiB. Optional language (a BCP 47 tag, or multi) sets the spoken language. Returns the transcript, the recognizer's confidence and each word's timing. Charged to the workspace's credits per second of audio; a workspace that cannot pay is refused with 402 before the audio is heard. Requests are rate-limited per person (429 with Retry-After).

_{Scope: run}