🎙️ a11y: Screen Reader Support for Dynamic Content Updates

[danny-avila]

Summary

I implemented initial screen reader support for dynamic content updates.

• Added a LiveAnnouncer component to manage and broadcast live announcements for screen readers.
• Implemented AnnouncerContext to provide announcement functionality throughout the application.
• Created utility components such as Announcer, MessageBlock, LiveMessage, and LiveMessenger to support screen reader announcements.
• Integrated the announcer functionality into the SSE event handlers for real-time updates.
• Uses semantically split chunks for processing of streamed content to provide more natural and coherent announcements (same strategy as text-to-speech audio stream).

Largely addresses:

• a11y: Updates to the chat are not announced for screen reader users #3570

Todo to close issue:

• Content updates for beginning/end of tool calls, image generation
• Content updates for Assistants API messages, i.e. content blocks with type "text".

Change Type

• New feature (non-breaking change which adds functionality)
• This change requires a documentation update

Testing

I tested the changes by generating AI responses and verifying that the screen reader announcements were triggered correctly. I also tested the streaming functionality to ensure that the announcements were coherent and timely.

Test Configuration:

• Screen reader: NVDA (latest version)
• Browser: Chrome (latest version)
• Operating System: Windows 11

Checklist

• My code adheres to this project's style guidelines
• I have performed a self-review of my own code
• I have commented in any complex areas of my code
• I have made pertinent documentation changes
• My changes do not introduce new warnings
• I have written tests demonstrating that my changes are effective or that my feature works
• Local unit tests pass with my changes