Repository navigation
Review Read Aloud - #2580
Review Read Aloud#2580
Conversation
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
There was a problem hiding this comment.
Read Aloud is approved and updated to your latest commit, @average-crumb. Nothing is required from you.
This submission matched an existing listing, so I treated it as a manual update. The listing now installs commit 42425c6 (the head of master, as in your submission). The code review found nothing to fix, and assets/screenshot.png still shows the speaker button under a reply, so the listing media stays as it was.
Your comment mentions pushing more fixes. They aren't in this review, since master still points at 42425c6. When you push more, open a new submission with the commit you want and I'll review that update.
Registry edits: moved the record from 4267479 to 42425c6, updated its review date, and removed the submission file. Your OVERVIEW.md is used as shipped.
Review data
- Artifact: git
https://github.com/average-crumb/paseo-read-aloudat42425c63022bed8806cfc04827fe4da4b200cd80(previously4267479fd7023f85d5f1e2d0b8f406c87c9c9ec7). Fetched at that exact commit without submodules or hooks; 21 files, no symlinks or submodules. Nothing was installed, built or run. - Compared old and new trees. Manifest and screenshot are unchanged. Changed:
client/,server/tts.ts,server/reply-row.ts,shared/,index.server.ts,OVERVIEW.md, version0.1.0→0.2.0, plus newtest/,README.md,LICENSE. - Install:
npm ci --omit=dev --ignore-scripts. Lockfile v3, 321 entries, allregistry.npmjs.orgwith integrity. The only production dependency issherpa-onnx@1.12.26(no scripts, no dependencies; integrity matches npm). The only install-script package,fsevents, is dev-only. The lockfile's dependency set is unchanged from the previous commit. - Network: one
fetch, to the OpenAI-compatible speech endpoint from Paseo's voice settings, and only when that provider is selected. It sends the spoken text with the configured API key and is triggered by a press or touch, never by hover. The 0.2.0 changes add cancellation and a timeout. - Credentials and files: reads Paseo's
config.json(voice settings, models directory, plugin path) and the matching voice environment variables. Reads voice model files from the models directory. Writes nothing to disk. The overview discloses the config read. - Execution: no subprocesses, no
eval, no remote code. Itrequires Paseo's ownsherpa-onnx-nodeand model catalog from the host install, and falls back to its own bundledsherpa-onnx. - Change in 0.2.0: sentence chunking, lookahead playback, cancellation and freeing the loaded voice. It adds no new network destinations, file access or execution.
- Not verified: behavior on a device (code read only).
- Validation: trusted registry validator on the updated head passed (
--online --changed).
Closes #2579. Submitted by @average-crumb.
Ready for artifact review. The reviewer will resolve the submitted source and complete the registry listing.
Submission:
submissions/2579.json.Git references are optional. Use a supplied reference; otherwise resolve main during review. Git updates are submitted manually.