- Introduced RunningHubVoxCPMDatasetBuildBatch class for constructing training manifests from a list of audio inputs, streamlining the dataset preparation process.
- Updated __init__.py to include the new batch dataset build node in NODE_CLASS_MAPPINGS and NODE_DISPLAY_NAME_MAPPINGS.
- Enhanced README_zh.md with detailed instructions for the new batch training workflow, including examples and parameter descriptions.
- Improved error handling in generate.py to ensure text inputs are validated as strings, enhancing robustness in audio generation workflows.
- Introduced RunningHubVoxCPMDatasetEntry and RunningHubVoxCPMDatasetBuild classes for creating and managing training datasets.
- Added RunningHubVoxCPMTrainLoRA and RunningHubVoxCPMTrainFull classes to facilitate LoRA and full model training directly within ComfyUI workflows.
- Updated NODE_CLASS_MAPPINGS and NODE_DISPLAY_NAME_MAPPINGS to include new training nodes.
- Enhanced README files with detailed instructions on using the new training functionalities and example workflows.
- Refactored generate.py and multi_speaker.py to integrate control instruction sanitization for improved text handling.
- Introduced RunningHubVoxCPMMultiSpeaker class for multi-speaker audio generation.
- Updated NODE_CLASS_MAPPINGS and NODE_DISPLAY_NAME_MAPPINGS to include multi-speaker functionality.
- Removed outdated example JSON files and added a new example for multi-speaker usage.
- Enhanced .gitignore to exclude sensitive files.
gluttony-10