Skip to content
Discussion options

You must be logged in to vote

An earlier discussion with some info about the normalization precomputed using Welfords algorithm - #1203 (comment)

The constants are precomputed for the dataset you wish to evaluate on. We computed on the then mixture of 3-4 datasets so Librispeech might be slightly different. The constants depend on several factors such as preprocessing window and stride so it is different between QuartzNet Jasper and the rest of the models (Citrinet, Conformer, ContextNet and Conformer Transducer)

Replies: 1 comment

Comment options

You must be logged in to vote
0 replies
Answer selected by roman-vygon
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Category
Q&A
Labels
None yet
2 participants