What problem does it solve? Raw speech-to-text output is a wall of unattributed sentences where nobody can tell who committed to what. This Skill turns meeting recordings into diarized transcripts with speaker labels and timestamps, so every decision, commitment, and quote is attributable and checkable against the recording. ## Core Features & Use Cases - Diarized transcription: Produces verbatim speech with segment-level timestamps and speaker labels (SPEAKER-01 and up, or known names only on voice match), with overlap marks for crosstalk. - Uncertainty handling: Flags low-confidence attributions as UNCERTAIN with timestamps and reasons instead of guessing speaker identity. - Grounded summaries: Closes with decisions, commitments, and open questions, each pinned to the timestamp that proves it. - Use Case: Hand the Skill a recorded interview or standup (m4a, mp3, wav, ogg) plus optional session context, and receive a dated markdown transcript with a speaker roster, timestamped segments, and a quotable summary. ## Quick Start Transcribe the attached meeting recording into a diarized transcript with speaker labels, timestamps, and a summary of decisions and action items.
