Skip to content
Sean Naren edited this page Nov 21, 2016 · 9 revisions

DeepSpeech.torch

The project aims to deliver end-to-end speech recognition based on deep neural nets in Torch7 trained using the Connectionist Temporal Classification (CTC) cost function. Both are heavily influenced and based on Baidus' DeepSpeech2 architecture and their open sourced version of CTC, Warp-CTC.

Pages to get started: