Jnewton-lab/JianYan
🎤 Transform speech to text on Windows with fast, local AI processing. Enjoy seamless recording and automatic integration for effective communication.
What's novel
🎤 Transform speech to text on Windows with fast, local AI processing. Enjoy seamless recording and automatic integration for effective communication.
Code Analysis
14 files read · 4 roundsA Windows system-tray voice-to-text tool that records microphone audio, transcribes it locally with FunASR/SenseVoice, optionally refines the transcript via an OpenAI-compatible LLM API, and pastes the result into the active application via simulated Ctrl+V.
Strengths
Well-structured Windows desktop app with thoughtful defensive programming (multiple clipboard/key-simulation fallbacks, single-instance guard via Win32 mutex, thread-safe state management) and a well-crafted LLM prompt for ASR post-processing. The PyInstaller packaging with custom hooks and Inno Setup installer shows production intent.
Weaknesses
Zero test coverage for a project with non-trivial threading, Win32 API calls, and audio processing. The README is generic with broken placeholder links and overstates features (e.g., 'Multi-Language Support' is just SenseVoice's auto-detect), and the 'fake progress' animation is a UX hack that doesn't reflect actual processing state.
Score Breakdown
Signal breakdown
Innovation
Craft
Traction
Scope
Evidence
Commits
15
Contributors
2
Files
50
Active weeks
4
Repository
Language
Python
Stars
1
Forks
0
License
MIT