search

louiskirsch / speechT

An opensource speech-to-text software written in tensorflow
Apache License 2.0
158 stars 36 forks source link

Language model while training #3

Closed arpit601 closed 7 years ago

arpit601 commented 7 years ago

Can we use Kenlm during training ?

louiskirsch commented 7 years ago

During training no language model is required. The language model is only useful during decoding. See https://arxiv.org/abs/1609.03193