Keep overlay visible while transcribing

[so1omon563]

Description

Keep the normal dictation overlay visible in its existing “Transcribing” state until final transcription finishes, then hide it through the shared post-output lifecycle.

The regression came from a normal-dictation-only branch that requested an asynchronous overlay hide before asr.stop() began. Removing that special case lets normal dictation use the same processing lifecycle already used by AI, Command, Rewrite, prompt-test, and sandbox routes.

Type of Change

• 🐞 Bug fix
• ✨ New feature
• 💥 Breaking change
• 🧹 Chore
• 📝 Documentation update

Related Issue or Discussion

Closes #676.

Testing

• Tested on Intel Mac
• Tested on Apple Silicon Mac
• Tested on macOS version: 26.5.2
• Ran linter locally: swiftlint --strict --config .swiftlint.yml Sources (SwiftLint is not installed locally)
• Ran formatter locally: swiftformat --config .swiftformat Sources (SwiftFormat is not installed locally)
• Ran tests locally: xcodebuild test -project Fluid.xcodeproj -scheme Fluid -destination 'platform=macOS,arch=arm64'

Also verified with swiftc -parse, git diff --check, an unsigned app build, and a real normal-dictation stop cycle. Runtime benchmarks confirmed processing=true before asr.stop(), the overlay remained visible while final transcription ran, and hide began only after asr.stop() returned.

Screenshots / Video

Before: reporter’s screen recording showing the overlay disappearing before transcription completes.

After: normal dictation keeps the overlay active through final processing and completes in the Test Playground. The final three seconds are a still of the completed result for readability.

• No UI/visual changes; screenshots/video are not applicable.

Notes

This removes 19 net lines and adds no new state or abstraction.

AI assistance: Codex was used for implementation and validation.

[greptile-apps]

Greptile Summary

The PR removes the normal-dictation early-hide branch so the existing “Transcribing” overlay remains visible until final transcription completes.

• Routes normal dictation through the shared processing-overlay lifecycle.
• Uses the shared empty-output and post-output hide behavior.

Confidence Score: 5/5

The PR appears safe to merge.

No blocking failure remains.

_{Reviews (3): Last reviewed commit: "Keep overlay visible while transcribing" | Re-trigger Greptile}

[altic-dev]

Hi is there a after video? I see the before video! Thank you! seems like a nice PR!

[github-actions]

This pull request has been marked stale because it has had no activity for 5 days. It will be closed in 2 days if there is no further activity. Add the keep-open or pinned label to prevent automatic closure.

[so1omon563]

Thanks! Here’s the after video:

FluidVoice-PR-793-after.mp4

This was captured against commit 161422b. It shows normal dictation keeping the notch overlay active through final processing, followed by the completed transcription in the Test Playground. The final three seconds are a still of the completed result so it remains readable.

[github-actions]

This pull request has been marked stale because it has had no activity for 5 days. It will be closed in 2 days if there is no further activity. Add the keep-open or pinned label to prevent automatic closure.

[so1omon563]

Rebased onto current main and resolved the conflict while preserving the new Spoken Send path. Local validation and all PR checks pass. Ready for review when you have time.