chunked/nltk_chunking/model_doc_speech-encoder-decoder/chunk_23.txt

Training
Once the model is created, it can be fine-tuned similar to BART, T5 or any other encoder-decoder model on a dataset of (speech, text) pairs.