mtsmfm.github.io

get_transcript

Return the whole session so far without moving the read cursor: every transcript segment ({t0,t1,text}, plus "corrected" — your correct_segments fix-up — where one exists; prefer corrected over text when it is there), every question ({id,t,text,answer,source,answeredT}; answer is null while unanswered, source is "spoken" | "typed" | "skipped", text is "" for a note the user typed without being asked), the glossary of proper nouns the user entered, notes (the Markdown memo shared with the user, from notes.md), language (the speech language whisper is transcribing in, "ja" or "en"), plus recording, ended and elapsed_s. Call this before writing the outline and the slides.

Page: /ramble-to-slides/

Input schema

No input properties (empty object).

{
  "type": "object",
  "properties": {}
}