PulseAugur
EN
LIVE 09:23:22

New REFRAMED dataset targets realistic audio description generation for movies

Researchers have introduced REFRAMED, a new dataset and formulation for generating realistic audio descriptions for movies. Unlike previous methods that relied on pre-specified content and timing, REFRAMED requires models to jointly decide what to describe and when to insert the descriptions into dialogue gaps. The dataset includes professional audio descriptions, subtitles, and aligned screenplays for 2,023 videos from 206 movies. While current state-of-the-art systems show improvement over basic baselines, they still fall short of human expert performance. AI

IMPACT Establishes a new benchmark for audio description generation, potentially improving accessibility tools for visually impaired audiences.

RANK_REASON The cluster describes a new academic paper introducing a dataset and a novel formulation for a specific AI task. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

New REFRAMED dataset targets realistic audio description generation for movies

COVERAGE [1]

  1. arXiv cs.CL TIER_1 English(EN) · Igor Sterner, Mirella Lapata, Alex Lascarides, Frank Keller ·

    REFRAMED: Towards Realistic Audio Description Generation for Movies

    arXiv:2608.09765v1 Announce Type: new Abstract: Audio Description (AD) is a verbal narration of key visual content in videos, enabling access for visually impaired audiences. Unlike standard video captioning, AD is a structured editorial task: descriptions must be inserted into g…