Live transcription preview while speaking
- Dominant language
- Swift
- Stars
- 2.9k
- Forks
- 226
- PR merge metrics
- No merged PRs in 30d
Description
**Is your feature request related to a problem? Please describe.**
Hex only shows transcribed text after I release the recording hotkey. While I'm dictating, I can't see what Hex is hearing, so longer utterances feel blind until I'm done talking.
**Describe the solution you'd like**
Show partial transcription **while the hotkey is held**, updating in real time even when the text changes or gets corrected mid-sentence. Final paste on release should remain authoritative.
Suggested UX:
- **Insert at cursor** — live preview in the frontmost text field
- **Overlay** — live preview in Hex's floating indicator without mutating the target app until release
**Describe alternatives you've considered**
- Overlay-only preview (no cursor mutation during recording)
- Streaming ASR vs periodic batch preview on growing audio snapshots
**Additional context**
- Parakeet models with Super Fast Mode capture path
- Electron editors (e.g. Cursor) may need keystroke/clipboard fallback when Accessibility insertion is unavailable
- Modifier-only hotkeys must not false-stop when preview keystrokes are injected
Contributor guide
No contributing guide indexed for this repository
Research direction
Start by tracing the existing recording hotkey, Super Fast Mode capture path, and release-time final paste; the payload names no files or tests. The preview strategy, insertion or overlay behavior, streaming approach, and Electron fallback still need decisions before completion can be defined.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- electron, macos, swift
- Domain
- audio-video-rtc, desktop
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Quiet
- Clarity
- Needs clarification
- Newbie friendliness
- 35/100