June 5, 2022
*Blog was created using Dan Povey’s interview with us: https://www.youtube.com/watch?v=TWgO-AfYuIE
If one is new to Kaldi it is recommended to start with LibriSpeech Model.
Out of those three there only one that would probably be accessible to students and beginners is LibriSpeech model because both Aspire and WSJ models rely on data from the linguistic data consortium.
LDC is not free unless you have a subscription which is expensive.
Dan Povey is recommending to start with Mini Libri Speech recipe which is like a small subset of LibriSpeech about three hours of training.
References:
#1 Which model to start with, youtube auto transcript