- Added AttrDict utility class for attribute-style access to dictionaries.
- Created voice encoder module with configuration and mel spectrogram processing.
- Developed VoiceEncoder class for embedding extraction from mel spectrograms.
- Introduced ChatterboxMultilingualTTS class for multilingual text-to-speech synthesis.
- Implemented conditionals handling for T3 and S3Gen models.
- Added functionality for generating speech from text with customizable parameters.
- Integrated voice conversion capabilities with ChatterboxVC class.
- Enhanced punctuation normalization function for better text input handling.
- Established loading mechanisms for model weights and configurations from local and pretrained sources.