File structure for train_lm and transcribe?

[Daniel Wolf]

Hi!

The documentation does a great job of describing the expected file structure for training an acoustic model. However, I couldn't find any information on the expected file structure for training a language model (mfa train_lm) and for transcription (mfa transcribe). The documentation merely mentions "the source directory to train from" and "the directory to transcribe", respectively.

Could someone tell me the expected file structure for these two commands?

Best,
Daniel