Spaces:
Running
Running
Commit History
-
fixing session mismatch
bebb6f7 -
fix references
55e07d1 -
Use sentence-based window sliding instead of word-based
- Fix complete sentences instead of individual words - Match Python implementation behavior (STT/smart_progressive_streaming.py) - Apply to both transcribeIncremental and transcribeBatch methods - Improves transcription quality by preserving sentence boundaries Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>c9d9124 -
Make UI more minimal and compact
- Remove "About This Demo" section for cleaner interface - Make controls thinner with reduced padding (p-6 → p-4) - Reduce label sizes (text-sm → text-xs) and spacing - Ensure all boxes (controls, transcription, metrics) are aligned with consistent max-width Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>b76aacc -
Fix update interval display from 250ms to 500ms in About section
Updated App.jsx to correctly show "Partial transcriptions every 500ms" instead of 250ms to match the actual implementation. Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>1b8c165 -
Fix performance metrics display and update intervals
- Add color coding for latency: green <500ms, yellow <1s, red >=1s - Update RTF color threshold: green >5x (was >10x) - Fix device display to show "WebGPU Hybrid" for webgpu-hybrid backend - Correct progressive update interval: 500ms (was incorrectly shown as 250ms) - Add descriptive text for webgpu-hybrid: "GPU encoder + WASM decoder" Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>f65234a -
Re-add progress bar functionality for model downloads
- Added progress callback to fromHub in worker.js - Progress tracking with initiate/progress/done messages - UI already has Progress component from earlier work - Note: Progress bars work locally but may not work on HF Spaces (investigating parakeet.js compatibility issue) Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>c54543a -
Add diagnostic logging to debug HF Spaces model loading
- Added console.log statements in worker.js to track fromHub execution - This will help diagnose why model loading is silent on HF Spaces - Locally shows [Hub] logs but HF Space shows nothing Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>f4b9755 -
Remove progress bar functionality and clean up debug logging
7438c56 -
Reduce timeout to 30s and add detailed CORS error message
ed2dc69 -
Add periodic progress check and reduce timeout to 60s for faster debugging
77ab1d4 -
Add IndexedDB logging and timeout detection for hung model loading
e4481a7 -
Add fetch interceptor to debug model download issues
82b8d7b -
Add debug logging to track model loading progress
cf635c3 -
Add dist files with progress bar support
0776db9 -
Fix progress bar display and correct model size to 2.5GB
- Convert parakeet.js progress format to UI-expected format - Send initiate/progress/done messages for progress bars - Track seen files to avoid duplicate initiate messages - Update model size from 2.1GB to 2.5GB (encoder is 2.3GB alone) - Progress bars now show download status correctly Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>2f7b75b -
Update model size from 2.5GB to 2.1GB
- Correct model size in Load Model button - Update footer info text with accurate size - Matches actual download size Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>7a0f76d -
Add Clear Cache button and caching info
- Add Clear Cache button next to Load Model button - Clear all IndexedDB databases when clicked - Add info text about model caching (~2.1GB cached locally) - Helps users understand and manage cached model files - Useful for testing and forcing fresh downloads Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>ab55401 -
Add download progress bar for model loading
- Create Progress component showing file download progress - Track initiate/progress/done messages from parakeet.js - Display progress bars with file name, percentage, and total size - Forward progress callback from fromHub to main thread - Show download status during model loading phase Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>86a84fc
-
Optimize performance with webgpu-hybrid and fix RTF display
- Use webgpu-hybrid backend (FP32 encoder on WebGPU + INT8 decoder on WASM) - Achieve RTF 25x+ (1.8x faster than parakeet.js demo) - Fix RTF calculation to show speed factor (higher is better) - Extend progressive streaming window to 15 seconds - Update UI: rename 'Audio Duration' to 'Window Size', remove 'Update Rate' - Performance: ~230ms for 5.8s audio (vs 411ms in demo) Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>ea6aed8 -
Final optimizations: VAD, 5s window, word-level sliding
- Simple amplitude-based VAD skips transcription during silence - Reduced window from 15s to 5s for faster sliding - Word-level fallback ensures window always slides at 5s - Removed excessive debug logging - Working with SharedArrayBuffer (12 threads) on HF Spaces! Performance: RTF 2-4x with encode times 400-1300ms (vs 5000ms single-threaded) Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>09b5ac6 -
Try COEP credentialless without COOP for SharedArrayBuffer
Using only COEP: credentialless should enable SharedArrayBuffer without breaking iframe embedding (no COOP required). Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>0e8835b -
Remove custom headers to fix iframe connection issue
The cross-origin headers were preventing the Space from loading in iframe. Will work with single-threaded WASM for now. Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>1251757 -
Add cross-origin headers and optimize progressive streaming
- Add _headers file with COEP/COOP for WebGPU support on HF Spaces - Implement simple amplitude-based VAD to skip transcription during silence - Reduce max window size from 15s to 5s for faster processing - Add word-level fallback when no sentence boundaries detected - Remove excessive debug logging for cleaner console output Fixes model loading hang on deployed Space due to missing cross-origin headers. Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>e74c555 -
Add microphone selector, audio level meter, and fix Chrome compatibility
- Add microphone device selector with auto-selection of default device - Add real-time audio level meter during recording - Fix stack overflow error in worker.js for large audio buffers - Reorganize UI to combine controls in one panel - Disable echo cancellation/noise suppression for Chrome compatibility - Add audio resampling from native rate to 16kHz - Include microphone test page for debugging Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>5151e54 -
Restructure for HuggingFace Space compatibility
- Move built files (index.html, assets/) to root for demo to work - Move source code to source/ folder for visibility - Clean file organization: demo works + source is browsable Structure: - Root: index.html, assets/ (working demo) - source/: all source code, configs, docs - README.md: Space documentation Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>0f739b8