What problem does it solve?
Manually splitting hour-long or multi-hour conference livestreams into individual per-talk transcripts, cleaning up caption errors, and adding structured timestamps and key takeaways is extremely time-consuming and error-prone for event organizers and content teams.
Core Features & Use Cases
- Automated Talk Splitting: Parses chapter timestamps from YouTube video metadata or description to automatically split long conference recordings into individual talk segments.
- Multi-Backend Transcription: Supports YouTube auto-captions (fastest, free option), Groq Whisper API, and local Whisper models with automatic fallback logic for different system setups.
- LLM-Powered Cleanup: Runs a cleanup pass to fix transcription errors, add key takeaways, format timestamps, and structure transcripts for readability.
- Use Case: For a 6-hour multi-talk conference livestream, this skill automatically generates 12 separate polished transcript files with speaker names, talk titles, and key points, eliminating hours of manual work.
Quick Start
Use the conference-transcribe skill to turn a multi-talk YouTube conference livestream URL into separate cleaned per-talk transcript files with key takeaways and timestamps.
