- Use singleton pattern from TTSService for efficient model management
- Remove complex manual memory cleanup code
- Simplify CLI arguments by removing redundant memory management options
- Load model once at start, let singleton handle efficient reuse
- Remove keep-model-loaded and cleanup-interval options
- Streamline generation logic to match backend service patterns
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Add audiobook generation CLI tool from dev branch
- Create import_helper.py to resolve backend service imports for CLI scripts
- All dependencies verified and working correctly
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Implement save/load buttons in dialog editor interface
- Add JSONL export/import for dialog scripts with validation
- Include timestamp-based filenames for saved scripts
- Add comprehensive error handling and user confirmations
- Create CLAUDE.md with development guidance and architecture overview
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Added aggressive memory cleanup for both single and dialog generation
- Implemented configurable silence gaps between dialog lines
- Organized output files into single_output/ and dialog_output/ directories
- Fixed torch import scoping issues
- Updated README with comprehensive documentation
Steve White