Repository navigation
Conversation
Sometimes on mobile it is not convenient to read a reply, especially a long one. Add a play button that uses the browser's native speech to read the reply aloud. Convert Markdown into speech-friendly text first, preserving prose and short identifiers whilst announcing omitted code blocks, diagrams, and large tables. Split long replies into smaller chunks for playback. Let users choose a voice and speaking rate in Settings. Keep Stop visible during playback, and stop reading when another reply is selected or the chat is closed.
Owner
Author
|
The screenshots and interaction clip now accompany powerfooI#505. This temporary evidence draft is therefore closed; the feature branch is retained for upstream review. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
chat-controls.mp4
The screenshot and silent clip show a genuine Pi reply after reading the public Northstar demo. The clip shows controls appearing on hover, hiding when the pointer leaves, and appearing again with keyboard focus. It demonstrates the UI, not audible playback. Captured from 90385be in Chromium 156 against a real Herdr backend; desktop viewport 1100 x 860, screenshot DPR 2, video DPR 1.
Summary
Sometimes a long agent reply is easier to hear than read, especially on a phone. This adds a small Speak button beside Copy in Agent History Chat, using the browser's native speech synthesis.
markedlexer: retain prose, short identifiers, link labels, image descriptions, and small tables; announce omitted code blocks, diagrams, and oversized tables instead of reading their syntax.There are no new dependencies, server endpoints, or external speech integrations. The formatter remains separate from playback. Existing asset budgets are unchanged.
Compatibility and limitations
Speech depends on the browser and its installed voices. Unsupported browsers show a disabled control. Some voices use online services and may send reply text to their provider; Settings and the feature documentation state this. Background and locked-screen playback are not guaranteed. No microphone permission is required.
Verification
bun run precommit— formatting, Oxlint, TypeScript checks, frontend build, and the complete test suite: 3,929 passed, 4 optional tests skipped, 0 failed.bun run typecheck:bun— all three projects passed.bun run build:web— passed, including asset budgets.bun run build:darwin-arm64.Oxlint reports one pre-existing
TerminalView.tsx:689warning. Vite also reports its existing large-chunk advisory; neither check fails.Speech settings
Mobile layout
Chromium viewport 390 x 844, DPR 2; this establishes layout only, not physical-device playback or keyboard behaviour.