Le Scribe Audio is a two-pass AI audio and video transcription platform designed for researchers, sociologists, journalists, debate writers, secretariats, municipalities, and CSE/CSSCT councils. It converts interviews, public debates, conferences, and meetings into structured transcripts with speaker diarization and timestamps, cutting manual drafting time in half (2 to 3 hours of full human re-listening and editing per audio hour instead of 4 to 7 hours of manual typing from scratch). Inside the synced interactive audio player, users can listen while editing, globally rename and code speaker diarization labels across every turn (for example replacing M. Martin with Enquete P1 or Intervenant A), generate an on-demand AI summary consisting of an executive summary followed by a chrono-thematic breakdown, and export a standard 100% editable Microsoft Word (.docx) document, PDF, SRT/VTT subtitles, or IRaMuTeQ/NVivo/ATLAS.ti corpora. Audio files and transcripts are stored in France on OVHcloud (AES-256 encrypted) with 1-click permanent deletion, while AI inference runs via Google Cloud Vertex AI (Europe region) with Zero Data Retention and no training on user files. Every new account includes 60 free trial minutes, and optional human proofreading is available at 60 EUR excl. VAT per audio hour by two successive transcribers.
