DiscourseKit

DiscourseKit is a web app (with optional desktop uploader) that converts raw audio/transcripts from meetings, interviews, podcasts, or fieldwork into structured discourse and conversation-analysis layers: speaker turns, overlaps, repairs, discourse markers, stance, hedges, and pragmatic functions. It’s an AI-assisted annotation tool, but it’s brutally practical: it focuses on producing exportable, auditable annotations rather than “insights” dashboards. Users can upload a transcript (or audio), get an initial auto-annotation pass, then correct it in an editor designed for linguists and qualitative researchers. Exports support common research workflows (ELAN, CSV/JSON, Praat TextGrid). The app also generates a transparent “annotation confidence” report so researchers can justify what was human-verified vs. machine-suggested. Realistically, it will not replace expert annotators; it will cut the boring first-pass work and standardize outputs across teams.