Mira: Re-imagining Web Accessibility through Context-Aware Audio Narration

Mira is a chrome extension that uses on-device AI (Gemini Nano) to transform fragmented webpages into coherent, context-aware audio narratives.

It moves beyond traditional "read-aloud" tools by focusing on intelligibility, not just audibility. Instead of linearly reading text and isolated alt-text, Mira analyzes the semantic role of visual media, weaving its meaning directly into the narration. This preserves the listener's mental flow and makes the web truly understandable as a coherent story.

✨ Features

• 🧠 Context-Aware Narration: Mira doesn't just read alt-text. It uses Chrome's built-in Gemini Prompt and Summarizer APIs to understand an image's purpose (e.g., as evidence, contrast, or an example) and integrates its meaning fluently into the surrounding text.
• 🎧 Coherent Listening Experience: By preserving the narrative flow, Mira reduces the "Split-Attention Effect" and lowers cognitive load, making web content as intelligible to listen to as it is to read.
• 🔒 On-Device & Private: Runs within the browser using Chrome's built-in AI stack. No user data ever leaves your machine.
• 🔊 Natural, Unbroken Audio: Leverages asynchronous chrome.tts buffering to create a seamless, pseudo-streaming narration that feels natural and fluent, without disjointed pauses.
• 🌍 Universal Design: Built for everyone—from those who need it for accessibility (e.g., blind or low-vision users) to those who prefer it for multitasking, learning, and audio-first consumption.

📋 Prerequisites

• Chrome version >= 138
• Get a Google ai studio api key and input in setting page.
• Hardware (from here)
  • Operating system: Windows 10 or 11; macOS 13+ (Ventura and onwards); Linux; or ChromeOS (from Platform 16389.0.0 and onwards) on Chromebook Plus devices. Chrome for Android, iOS, and ChromeOS on non-Chromebook Plus devices are not yet supported by the APIs which use Gemini Nano.
  • Storage: At least 22 GB of free space on the volume that contains your Chrome profile.Built-in models should be significantly smaller. The exact size may vary slightly with updates.
  • GPU or CPU: Built-in models can run with GPU or CPU.
    • GPU: Strictly more than 4 GB of VRAM.
    • CPU: 16 GB of RAM or more and 4 CPU cores or more.
  • Network: Unlimited data or an unmetered connection.

🛠️ Installation

• Open chrome and go to chrome://flags/#prompt-api-for-gemini-nano and chrome://flags/#prompt-api-for-gemini-nano-multimodal-input, enabling both flags.

• Build the unpacked extension

  • Build from source: npm install and npm run build to generate the dist directory.

  • Or get the built package from release (dist.zip) and extract it.

• Load it in chrome://extensions/ after enabling Developer mode

• Input your Google AI Studio API key in the settings page

💡 Tips

• Reading the page at first may be slow, because it is generating image caption (See in F12 developer panel)
• Previous tab's image caption generation process is killed when switching tabs, so don't worry for high CPU usage!