Towards the Usage of Pauses in Audio-Described Videos
Benoît Encelle, Magali Ollagnier Beldame, Yannick Prié · 2013 · Proceedings of the 10th International Cross-Disciplinary Conference on Web Accessibility (W4A) · doi:10.1145/2461121.2461130
Summary
This paper explores the use of "artificial pauses" — brief interruptions inserted into video playback — as a technique for delivering audio descriptions that cannot fit within the natural gaps in a video's soundtrack. Classical audio description is constrained by the duration of silence between dialogues or important sound elements; when gaps are too short, describers must omit key visual information. The authors propose inserting pauses immediately after natural gaps to extend the available time for descriptions. They tested this approach with 18 legally blind participants (visual acuity less than 1/50) who watched an enriched version of a nearly unknown 9-minute cartoon called "Tong." The researchers identified 41 key visual elements requiring description, of which 26 could not fit within natural gaps and required artificial pauses. Three pause durations were tested: D1 (half the gap duration), D2 (equal to the gap duration), and D3 (1.5 times the gap duration). Participants watched their assigned version twice and pressed a key whenever they felt discomfort from a pause, providing both quantitative discomfort data and qualitative feedback via questionnaires.
Key findings
Pause duration had no statistically significant impact on perceived discomfort (chi-square p=0.72), meaning even the longest pauses (1.5x gap duration) were no more disturbing than shorter ones. However, pause location within the video did significantly affect discomfort (p<0.006), with earlier pauses causing more discomfort — particularly around the 4th pause — suggesting an adaptation effect as viewers became accustomed to the technique. Critically, pauses were significantly less disturbing on the second viewing than the first (p<0.009), confirming that users habituate to artificial pauses over time. Qualitative results were strongly positive: all 18 participants found the experiment pleasant, all were self-motivated, 14 of 15 found the story comprehensible, and none found it annoying. Participants judged pauses as more helpful than disturbing. They suggested pauses would be especially valuable for fast-paced content like action movies, and could also convey visual-only information such as on-screen text and flashbacks. Some participants recommended adding a brief audio cue before pauses to make them less surprising.
Relevance
This research addresses a real and persistent limitation of audio description: the constraint of fitting descriptions into natural silences means that visually complex or dialogue-heavy content often receives incomplete description. The finding that artificial pauses are well-accepted by blind users — and that acceptance improves with exposure — opens a practical path for richer audio description of content that would otherwise be under-described. For video platforms and content creators, this suggests that extended audio description (pausing the video to insert descriptions) is a viable technique that should be offered alongside inline description. The participant suggestion of adding audio cues before pauses is a useful design recommendation. The study is limited by its small sample size (18 participants) and use of a single cartoon rather than diverse video genres, but it provides encouraging evidence for a technique that has since been recognized in standards like WCAG as "extended audio description."
Tags: audio description · video accessibility · blindness · visual impairment · multimedia accessibility · speech synthesis