Speech Transcript Summaries

Shows every persisted flow summary, one card per round — a concise per-argument summary of that round's flow, plus suggested cross-examination questions and extension ideas for anything still unanswered.

  • Route: /summaries
  • Package: debate-round

What it shows

Each persisted FlowSummaryRecord renders as its own card with the round's per-argument summary and, for unanswered arguments, ranked cross-exam questions and extension ideas (strongest opportunity first). A "Generate from raw speech text" form lets AI derive a round's summary from one or more pasted transcripts, with microphone dictation support and bulk multi-speech upload.

Data flow

state/flowSummaries.ts (localStorage: flowSummaries)
  → buildFlowSummariesPanelView()
  → panels/FlowSummariesPanel.tsx
  → apps/debate-ai.com/app/summaries/page.tsx

Generating from raw text:
  → extractTranscriptsBulk(...) — round/bulk-transcript-extraction.ts
      → requestTranscriptExtraction() — round/transcript-extraction-client.ts
          → POST /api/reason-ai

Known gaps

Microphone dictation transcribes live speech only — no already-recorded audio/video file upload.