How jasoncheng7115/jt-live-whisper works
jt-live-whisper is a local-first speech toolkit for live captions and translation, offline audio transcription, diarization, meeting analysis, and speech synthesis. The main user workflow captures system audio or accepts a recording, transcribes and optionally translates it, then presents captions or produces transcript and meeting outputs. A separate REST API exposes queued transcription jobs, persistent results, and webhook events; WebUI provides browser-based controls. The graph includes documented capabilities even where implementation was not sampled. Arrows are limited to README statements and source-index references; unsampled internals are not inferred.
Main entry points: translate_meeting.py Transcription Pipeline · app.py Job API · engine.py Job Processing Engine · webui.py Browser WebUI
Components 21
Capture and Transcription
- System Audio Capture
sck_audio_capture.swift - Transcription Pipeline
translate_meeting.py - GPU Speech Server
remote_whisper_server.py - Offline Audio Processing
translate_meeting.py
Meeting Outputs
- Speaker Diarization
translate_meeting.py - Transcript Parsing
jtdt_meeting/transcript_parse.py - Meeting Analysis
jtdt_meeting/meeting_insight.py - Speaker Charts
jtdt_meeting/meeting_charts.py
REST API
- Job API
jtlw_api/app.py - API Configuration
jtlw_api/config.py - Job Processing Engine
jtlw_api/engine.py - Job and Result Store
jtlw_api/store.py - Events and Webhooks
jtlw_api/events.py
Speech Synthesis
- Speech Synthesis
jtlw_tts/engine.py - Taiwanese Reading Rules
jtlw_tts/tw_reading.py - Voice Interpretation
jtlw_tts/interp.py - Interpreter Microphone
jtlw_tts/vmic.py
User Interfaces
- Subtitle Overlay
subtitle_overlay.py - Browser WebUI
webui.py
Other components
- User
- API Client
Connections 21
- User → System Audio Capture: provides audio
README.md - System Audio Capture → Transcription Pipeline: feeds audio
- User → Offline Audio Processing: submits recording
README.md - Transcription Pipeline → Subtitle Overlay: displays captions
README.md - Browser WebUI → Transcription Pipeline: starts process
webui.py - Transcription Pipeline → GPU Speech Server: uses remote ASR
README.md - Offline Audio Processing → Speaker Diarization: supports diarization
README.md - Offline Audio Processing → Transcript Parsing: parses transcript
- Offline Audio Processing → Meeting Analysis: produces analysis
README.md - Meeting Analysis → Speaker Charts: includes statistics
- API Client → Job API: submits jobs
README.md - Job API → API Configuration: loads settings
jtlw_api/app.py - Job API → Job Processing Engine: dispatches jobs
jtlw_api/app.py - Job API → Job and Result Store: persists jobs
jtlw_api/app.py - Job Processing Engine → Job and Result Store: saves progress
jtlw_api/engine.py - Job Processing Engine → Events and Webhooks: emits progress
jtlw_api/engine.py - Job Processing Engine → Transcription Pipeline: uses speech functions
jtlw_api/engine.py - Browser WebUI → Speech Synthesis: uses synthesis
webui.py - Speech Synthesis → Taiwanese Reading Rules: applies reading rules
jtlw_tts/engine.py - User → Voice Interpretation: uses interpretation
README.md - Voice Interpretation → Interpreter Microphone: sends speech