Espnet: Is it possible to reuse Librispeech LM without wordpieces model?

[smolendawid]

I'd like to use the LM model from here:

https://github.com/espnet/espnet/blob/master/egs/librispeech/asr1/RESULTS.md#pytorch-large-transformer-with-specaug-4-gpus--large-lm

Unfortunately, a word-pieces model isn't attached, so I am not able to train my own acoustic part, because I don't have unified inputs. Am I right? Is anyone able to share the correct BPE model with me?

[sw005320]

@ShigekiKarita or @kan-bayashi, could you upload the BPE model?

[kan-bayashi]

Here you are.
https://drive.google.com/drive/folders/1eO9uLEW0c6DlmpmSU8PdmTJDdJpyC5jp?usp=sharing

[sw005320]

Thanks a lot!

[smolendawid]

Thank you!