Automate video chaptering with LLMs and TF-IDF: Transform raw transcripts into well-structured documents
-
Updated
Sep 9, 2024 - Jupyter Notebook
Automate video chaptering with LLMs and TF-IDF: Transform raw transcripts into well-structured documents
Evaluate chaptering quality for audio and video content in time space, supporting segmentation and title generation
Command line utility to manage a lightweight markup writing (markdown for example) split in chapter files.
Podcast chapterization with LoRA-fine-tuned Qwen3-4B-Instruct: transcripts → hierarchical table of contents. End-to-end pipeline with Whisper ASR, ToC-LLM distillation, two-stage duration-specific fine-tuning, and WindowDiff/ROUGE-L/SBERT evaluation on YTSEG.
To associate your repository with the chapterization topic, visit your repo's landing page and select "manage topics."