3.5 / 5

Screen Narrator

Describes what's on screen for accessibility or note-taking

Serious

Privacy

Instant

The Problem

Screen readers miss context; manual note-taking during video calls is distracting

Current Solutions (Not Great)

Basic screen readers (miss visual context), manual notes (distracting), cloud tools (privacy risk)

Who Needs This

Visually impaired users, students in online lectures, professionals on video calls

Screen Narrator captures what's on your screen and provides real-time AI descriptions. For visually impaired users, it describes UI elements, images, and layouts that screen readers miss. For sighted users, it can generate notes from video calls, lectures, or tutorials. Because inference is local, your screen content—which might include sensitive work—never leaves your device.

Honest Take

Noble accessibility goal with real impact potential. But screen readers are complex assistive tech with decades of development. The meeting notes angle is more immediately useful. Privacy is a genuine differentiator for work screens. Consider partnering with accessibility organizations for feedback and distribution.

Monetization Ideas

Ways to turn this into revenue

Freemium

Free basic, $5-15/mo for pro

Subscription

$5-29/month or $49-199/year

Donations / Tip Jar

$5-50/month from grateful users

Features

Key features that make this app valuable

Real-time screen description at configurable intervals
Focus on changes only (doesn't repeat static content)
Audio output via speech synthesis
Transcript log for review
Selective region capture
Meeting notes mode (captures slides + speaker)
Customizable verbosity
Hotkey activation

Build Prompt

Use this prompt with an AI assistant to start building

Build a React PWA called 'Screen Narrator' using WebInfer with vision capabilities and Screen Capture API. UI: 'Start Narration' button, live preview thumbnail, scrolling transcript log, verbosity slider. Use getDisplayMedia() to capture screen. Capture frame every 2-5 seconds (user configurable). Compare frames using canvas pixel diff to detect significant changes. On change, send frame to vision model: generateObject returns { description: string, keyElements: string[], textContent: string[], changes: string[] }. Use speechSynthesis API to read description. Log all descriptions with timestamps to scrolling transcript. 'Meeting Mode' adds speaker detection and slide change alerts. Export transcript as markdown. Hotkey to pause/resume.

Score Breakdown

Pain