Feat(server): support WHISPER_MODEL_NAME/WHISPER_MODEL_SIZE (e.g. tiny); prefer local <name>.pt; improve error guidance 88e4729 rjzevallos commited on Nov 19, 2025
Fix(app): init ASR model in background; add error handling; remove share=True bcc3253 rjzevallos commited on Nov 19, 2025
Fix: send 'FINISH' text over WebSocket on stop to match server 921076a rjzevallos commited on Nov 19, 2025
Fix: send 'FINISH' text over WebSocket on stop to match server 408e16b rjzevallos commited on Nov 19, 2025
Fix: send 'FINISH' text over WebSocket on stop to match server 89b1173 rjzevallos commited on Nov 19, 2025
Fix: send 'FINISH' text over WebSocket on stop to match server 20547d7 rjzevallos commited on Nov 19, 2025
WS: support binary finish marker; client sends 0xFF.. marker (client only sends binary chunks) 4ae7ed6 rjzevallos commited on Nov 19, 2025
WS: accept both binary audio chunks and text control messages (FINISH/RESET) b809819 rjzevallos commited on Nov 19, 2025
Fix: send 'FINISH' text over WebSocket on stop to match server d711f84 rjzevallos commited on Nov 19, 2025
UI: Add visible Start/Stop recording buttons with improved layout 071c394 rjzevallos commited on Nov 19, 2025
Implement true streaming: WebSocket for real-time chunk processing 1e30c31 rjzevallos commited on Nov 18, 2025
Refactor: Use native Gradio audio component for reliable microphone capture 0eb9991 rjzevallos commited on Nov 18, 2025
Fix: Optimize build to prevent OOM - download model at runtime instead of build time 3e8e10c rjzevallos commited on Nov 18, 2025
Improve UI: Fix audio recording with better button detection and feedback f8ce589 rjzevallos commited on Nov 18, 2025
Fix: Use openai-whisper library for model loading instead of torch.hub 3b07c9e rjzevallos commited on Nov 18, 2025
Add simulstreaming_whisper module, update requirements, improve Dockerfile and model handling d860e14 rjzevallos commited on Nov 18, 2025
