This is an open source project (formerly named Listen, Attend and Spell - PyTorch Implementation) for end-to-end ASR implemented with Pytorch, the well known deep learning toolkit.
Hi, when representing the text as 'subword' feature, we need pretrain a text encoder, just like that in preprocess_libri.py. But the training step is missed in preprocess_timit.py, so an error will occur when we represent text as 'subword' feature for timit dataset, that is "Key error: ". So @Alexander-H-Liu can you fix this error if you have time? Thank you very much!
Hi, when representing the text as 'subword' feature, we need pretrain a text encoder, just like that in preprocess_libri.py. But the training step is missed in preprocess_timit.py, so an error will occur when we represent text as 'subword' feature for timit dataset, that is "Key error:". So @Alexander-H-Liu can you fix this error if you have time? Thank you very much!