Sky Glass TV Voice Control Design

1 July 2025

Role:

Led design and definition of meeting accessibility standards according to W3C
Applied lived experience to advocate for practical, inclusive changes
Led user interviews
Co-led final stakeholder presentation
Helped adapt team structure to better handle time zone conflicts
Contributed to critiques, maintaining accessibility principles throughout

Designing Inclusive Voice Control for Shared Home Use

Sky Glass redesigned homescreen with voice guidance shown mockup shown on sky glass tv

Summary (speed-read approved)

Sky’s voice control system wasn’t meeting the needs of real households with accessibility and personalisation requirements. Our team redesigned the voice setup flow to support multiple users with individual accessibility preferences, making the experience more inclusive, intuitive, and user-led. I led the accessibility research and design focus, ensuring user needs were addressed in every stage of the process.

The Challenge

As part of my training in my recent level 7 qualification UX course with KCL, Sky was an industry client who commissioned teams of students to redesign the voice control system of their Sky Glass TVs.

They made it clear that accessibility was up most in their minds, and wanted something innovative, ‘blue sky thinking’. Taking this on board, we came up with the following concept. Sky Glass TV’s voice interaction system was designed for single-user control — but in real homes, TVs are shared by families with vastly different needs. Our research showed that users were frustrated by inaccessible defaults, inconsistent voice control, limited customisation, and a lack of visual feedback. This issue was one I raised due to personal experience with members of my household who have mixed accessibility needs, and was met with enthusiasm by my team and Sky.

One of our hypothetical user scenarios based on real user interviews captured this clearly:

The Sharma Family

Zayan (8) is sensitive to noise and light
His mum struggles with hearing but avoids subtitles
His granddad has arthritis and struggles with the remote

They constantly had to re-adjust the TV for each person — a frustrating, inaccessible experience.

Cartoon Family Portrate with scenario solutions summarised

Goals

Create user-specific voice profiles that store individual accessibility preferences
Reduce friction in shared households
Improve clarity, trust, and control in the voice interaction setup

Constraints

Interface had to remain simple, friendly, and voice-first
Had to be usable without a touchscreen or physical input
Voice data privacy was a major user concern

We had 4 weeks to complete this project with the goal to have a mid fidelity prototype to show our ideas by the end of this timeframe.

Research

We used a combination of interviews, competitor research, and empathy mapping to uncover key pain points with current voice control systems.

Research Methods

Discovery interviews with users aged 20–60. Participants came from diverse cultural and national backgrounds including India, the UK, and other parts of Europe. Interviews were conducted with users who had wide range of accessibility issues
Empathy maps based on real quotes and behaviours
Competitive analysis (Alexa, Siri, Samsung, Google TV)
Problem definition and “How might we…” workshop
Wizard of Oz testing to simulate voice interactions

Key Insights

Users wanted fast, non-intrusive control, especially when multitasking
Distrust was common due to lack of transparency and constant listening
Accessibility settings were hard to find or confusing
Voice feedback (beeps, icons) was inconsistent or missing
Users wanted more personalisation beyond functional commands

Research Images

Strategy

We reframed the problem as:

How might we improve Sky Glass to adapt to different users’ needs — effortlessly, enjoyably, and without being overwhelming?

We prioritised features that offered user control, clarity, and adaptability, grounded in UX heuristics:

Recognition over recall
Visibility of system status
User control and freedom
Flexibility and efficiency of use

Design Process

We created a mid-fidelity prototype walking users through voice-based profile setup. The TV could recognise users and automatically adjust key settings.

Key features:

Voice-activated wake-up (“Hello Sky”)
Profile creation flow with name, avatar, preferences
Interaction tone selection: Friendly, Neutral, Adaptive
Custom voice phrases (e.g. “Open the table” → “Open football scores”)
Smart accessibility setup: Visual, Audio, Subtitles, Sign Language, Mobility
Shared Viewing Mode that blends needs of all users

Initial Mid-fidelity Prototype Design

User Testing & Iteration

We tested using the Wizard of Oz method, with one interviewer asking the questions, and the other acting as the Voice control, replying in kind to the users (it was quite a fun experience with lots of laughs all around). We interviewed 8 people and gathered feedback from their interactions with our voice prototype.

What we learned:

Text readability was impacted by background images
Progress indicators were unclear or missed
Abstract labels like “Control Sky Glass entirely with your voice” were confusing
Avatar names like “Thunder” or “Rain” didn’t make immediate sense to users
Setting screens felt like reading a manual — too much at once
Users wanted previews before enabling accessibility features
Voice data privacy and control were serious concerns

How we improved it:

Softened backgrounds and improved contrast
Added step indicators
Split avatar/voice tone into separate steps, renamed avatars (e.g. Amira, Keith)
Added visible “Skip” button
Reduced categories from 6 to 5 and switched to horizontal carousel
Renamed “Voice Control” to “Voice Guidance” and added previews
Introduced sliders for screen reader volume/speed

Old vs new setting screen (reduction from 6 → 5 categories)
Use arrows, labels or subtle highlights to show the change clearly.

My Role & Contributions

Led design and definition of meeting accessibility standards according to the W3C
Applied lived experience to advocate for practical, inclusive changes
Led user interview prep, using background in psychology
Led user interviews
Co-led final stakeholder presentation
Helped adapt team structure to better handle time zone conflicts
Contributed to critiques, maintaining accessibility principles throughout

High-Fidelity Prototype

Impact and Takeaways (no not the deliveroo kind sadly)

This project reinforced that inclusive design is never one-size-fits-all — it must adapt to diverse bodies, minds, and moments.

Designing voice-first for a shared device meant thinking beyond features — it meant designing for family dynamics, cognitive load, and emotional friction.

Next Steps

Test updated prototype with users who have specific accessibility needs
Run co-design workshops with people with lived experience of disability
Explore passive profile recognition (e.g., voice pattern detection)

Tools Used

Esther Weedon