chunked/nltk_chunking/model_doc_clap/chunk_3.txt

The CLAP model uses a SWINTransformer to get audio features from a log-Mel spectrogram input, and a RoBERTa model to get text features.