Scriberr 0.4.0 Release
This PR merges the release/0.4.0 branch into main, bringing several major enhancements and stability improvements to the core application.
🎙️ Speaker Diarization
- Added full speaker diarization support using PyAnnote's speaker-diarization-3.1 model
- Enhanced word-level speaker labeling with improved accuracy
- Implemented HuggingFace token integration for model access
- Built-in fallback mechanisms when diarization encounters issues
- User-friendly speaker label editing in the UI
🔐 Authentication & Environment Variables
- Improved authentication system with token refresh and proper expiry handling
- Better runtime environment variable management
- Streamlined setup process with API key validation
- Enhanced error handling and user feedback during configuration
🖥️ UI Enhancements
- Updated AudioPlayer component with improved controls and visualization
- Better markdown rendering for summaries
- Support for "thinking" sections in LLM responses
- Enhanced mobile responsiveness across all components
- Improved file management interface
🐳 Docker & Deployment Improvements
- Added optimized CUDA 12.8 support for GPU acceleration
- Renamed and reorganized Docker workflow files for clarity
- Improved hardware detection and dependency installation
- Enhanced GitHub workflow automation for releases and versioning
- Better ARM64 architecture support
💾 Other Improvements
- Enhanced batch size configuration for Whisper model
- Improved GPU memory management for more efficient transcription
- Added fallback for GPUs without efficient float16 support
- Enhanced title generation with improved system prompts
- Better error logging throughout the application
Testing
This release has undergone extensive testing in the following areas:
- Speaker diarization with various audio files and speaker counts
- Authentication flows and environment configuration
- UI rendering on different devices and screen sizes
- Docker deployment on both CPU and GPU environments
- Performance optimization for large audio files
🤖 Assisted By [Claude Code](https://claude.ai/code
Shortened Changelog
- feat: enhance Docker setup and transcription features by @SpirusNox in #47
- Optimize Docker setup and improve transcription handling by @SpirusNox in #49
- GPU Support, Docker Updates, Tailwindcss revert, Tailwindcss fix after revert, Readme update by @SpirusNox in #52
- Update GitHub Actions workflow for Docker image builds by @SpirusNox in #59
- Update GPU configuration, Docker setup, Fixed Arm64 build error by @SpirusNox in #60
- Improvement/update GitHub workflows by @SpirusNox in #64
- feat: Update Dockerfiles and move dependency installations to deployment by @SpirusNox in #63
- Docker update workflow updates by @SpirusNox in #65
- Rename Docker workflow files to reflect CUDA context by @SpirusNox in #67
- Feature: Handle thinking field support and fix markdown rendering by @SpirusNox in #72
- Revert "Feature: Handle thinking field support and fix markdown rendering" by @SpirusNox in #73
- fix: improve authentication with token refresh and expiry handling by @SpirusNox in #70
- feat: enhance audio quality for playback by @SpirusNox in #71
- Fix runtime environment handling and speaker diarization by @SpirusNox in #75
- feat: Add batch size configuration for Whisper model and improve diarization logging by @SpirusNox in #76
- fix: Update batch size handling for transcription and improve documentation on GPU memory issues by @SpirusNox in #77
- Fix/diarization process by @SpirusNox in #79
- Merging of nightly to 0.4.0 by @SpirusNox in #80
- Fix/diarization process by @SpirusNox in #81
- fix: Diarization fix by @SpirusNox in #82
- feat: Update GitHub workflows for automated releases and versioning by @SpirusNox in #88
- fix: Add fallback from float16 to float32 for GPUs without efficient float16 support by @SpirusNox in #89
- fix: Improve CUDA memory management for diarization by @SpirusNox in #90
- fix: UI enhancements by @SpirusNox in #91
- Update release/0.4.0 with latest nightly changes by @SpirusNox in #92
- Release version 0.4.0 to main by @SpirusNox in #93
- Merge nightly branch into release/0.4.0 by @SpirusNox in #94
- Release version 0.4.0 to main by @SpirusNox in #95
- Update README.md by @SpirusNox in #97
Full Changelog: 0.3.0...v0.4.0