Changes
- Lexicon validator in PER algorithm
- Generalize Transcriber to any-level offsets (word, character/phoneme, sentence)
- Refactored Submodules
- Extend customizability of Runner
- Implemented Word-overlap Segmentation
- Removed batched audio classification
- Output dataset to HuggingFace Dataset utility script
- Added guides, demos, examples