Listen, Reason, and Segment: Aligning LALMs with Editorial Judgment for Media Chapterization

Large Audio Language Models (LALMs) have made rapid progress on standardized benchmarks, yet their deployment in practical media workflows, curation, archival indexing, and content distribution remains largely unrealized. We identify automated audio chapterization, the task of segmenting continuous ...