Description

A study exploring how audiences personalise subtitle visibility in real-time. Through survey analysis and prototype evaluation, the research identifies how adaptive, speaker-selective subtitles can improve engagement and reduce distraction.

Adaptive Subtitles – Real-Time Control for Better Viewing Experiences

Subtitles are widely used to improve comprehension of media. They support language learning, assist people with hearing differences, and counter noisy viewing environments. Yet, subtitles can also divide attention. Some viewers report distraction, reduced immersion, or fatigue when reading while watching — suggesting that subtitles work best when they match individual preference, context, and cognitive load.

This project investigated how audiences currently use subtitles and what challenges they face. The research focused particularly on the idea of personalisation, exploring how giving viewers real-time control over subtitle visibility — down to the level of individual speakers — might improve both accessibility and enjoyment.

Why Explore Personalised Subtitle Control?

Viewers differ in how they consume media:

• some need constant subtitles for clarity
• others only use them selectively
• many switch them on or off depending on environment
• some prefer visual focus over reading
• multilingual viewers may need subtitles for certain characters or accents

Traditional subtitle systems offer binary choice: all subtitles or none. We hypothesised that more granular control could create a better balance between readability and immersion.

Study Design

The research was conducted in two main phases:

1. Online Questionnaire – 102 Participants

The survey explored:

• when and why people enable subtitles
• perceived benefits in comprehension
• perceived drawbacks including distraction
• preferences for control and customisation

These findings formed the design requirements for a new adaptive approach.

2. Prototype Evaluation – 19 Participants

We then developed Adaptive Subtitles, a real-time interface where viewers can enable or disable subtitles on a per-speaker basis while watching. Participants used the prototype to explore:

• interaction trade-offs
• cognitive load vs clarity
• comfort and usability over time

The study assessed whether dynamic control helped viewers feel more in command of their viewing experience.

Key Findings

The evaluation revealed several insights.

• Personalisation improved viewing enjoyment
  Participants reported that being able to limit subtitles to specific characters reduced screen clutter and helped focus attention.

• Granular control supported cognitive balance
  Viewers toggled subtitles strategically — enabling them for complex dialogue, unfamiliar accents, or key plot revelations.

• Structured onboarding was crucial
  Clear guidance helped viewers understand the benefit of selective control and prevented overload from too many options.

• Trade-offs were accepted when transparent
  When the system made benefits and limitations visible, participants were more confident in adapting subtitles dynamically.

Overall, the study demonstrated that real-time adaptive subtitles can improve viewing experiences when supported by thoughtful interaction design and guided user onboarding.

Implications & Future Work

This research suggests new directions for media accessibility and content personalisation:

• context-aware subtitles that adapt automatically
• profiles for focus, accessibility, education, or multi-language viewing
• real-time speaker highlighting for group scenes
• emotion-aware or scene-aware subtitle behaviour
• integration into streaming platforms and VP environments

Personalisation signals an evolution in how audiences interact with media — not just as passive viewers, but as agents tailoring content to their preferences, needs, and contexts.

Adaptive Subtitles contributes to ongoing research into viewer-centred accessibility design, emphasising control, clarity, and intentional choice as core elements of better media engagement.