Karthik Ganesan

Thank you

@chorongi can you add the TODO tasks here I will take care of this PR from here ?

@jhlee9010 thanks for this contribution . could you share a link to your trained model ? Also @b-flo is there any way we can decode a Non-streaming conformer-RNNT https://huggingface.co/espnet/chai_librispeech_asr_train_conformer-rnn_transducer_raw_en_bpe5000_sp via...

@b-flo yes I would like to know how to perform online decoding of non-streaming conformer https://huggingface.co/espnet/chai_librispeech_asr_train_conformer-rnn_transducer_raw_en_bpe5000_sp . Thanks for info on online transducer will get it in touch with Yifan

@jwrh afaik the issue arises from this line `utils/copy_data_dir.sh data/"$(echo "validated_${lang}" | tr - _)" data/${train_set}` Please try add some debugging in the above script to verify the same. @simpleoier...

@jwrh can you try working on a subset of data (e.g just dev in place of train) then we can quickly know if it is efficiency issue. As the same...

@jwrh from the number of utterances , the French dataset seems to be 6x in terms of number of utterances. please `htop` on another terminal / window , to see...

It's data copying issue , the French dataset is 5x bigger ,so leading to memory issues ,the same scripts works for Welsh as it's small

> Are you planning to upload a model to the Hugging Face Hub? > If so, please add a link to the model to README.md Sure we are currently trying...

@simpleoier please let us know if anything is left to be added , if not it would be great if you can merge the PR