SeanNaren / deepspeech.torch

Speech Recognition using DeepSpeech2 network and the CTC activation function.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Trained model file download

zavalyshyn opened this issue · comments

commented

Hi,

I've been running the Train.lua script for days on my setup (no GPU, just CPU) and managed to move to the second epoch.

Is there a way to download a pretrained deepspeech AN4 model from somewhere?

I'd really appreciate this. Thanks

Hey sorry for the late response. I'll do this when I can, I plan to release two models for Librispeech and for AN4, but not entirely sure where to put it, any suggestions?

EDIT: I'll make a release and put it there

commented

Hey, cool thanks.

I can store it on my Dropbox. I have 27 gigs there which I don't use.

You can also store it on my google drive if you are interested. I have unlimited space.

@SeanNaren Thank you Sean! What is the best WER% you have been able to achieve so far?

Sorry for the disappearance, everything can be seen here!

I realised some people may want a CPU only model (would take forever to train on Libri, I can do this for AN4 only). I'll make a ticket for this.

commented