LAS_Mandarin_PyTorch icon indicating copy to clipboard operation
LAS_Mandarin_PyTorch copied to clipboard
verified publisher icon jackaduma

How to train the LM?

Open lhang33 opened this issue 5 years ago • 3 comments

Hello again! I have modified your ASR model struture and used a larger madarin dataset to train it. The CER has already reached down to 10.5 after 1m step without Language Model.

I'd like to see how further it improves by using the LM. But I don't know how to train on your LM codes yet. Could you please tell us how to train it, infer with it, or share your "yaml" file to us? Thanks in advance!

lhang33 avatar Oct 10 '20 09:10 lhang33

to train the LM: 1.first replace the wavs with corpus 2.python3 main.py --config config/lm_example.yaml --lm

i hope these steps will help you

durian1023 avatar Mar 10 '21 03:03 durian1023

Would like to ask you have met the problem "CUDNN_STATUS_NOT_INITIALIZED" while train the ASR with CTC? Thanks in advance!

durian1023 avatar Mar 10 '21 03:03 durian1023

hi,can you please teach me how to convert TIMIT data to JSON file, so that I can run the model.Thanks!

comingway avatar Mar 20 '21 14:03 comingway