Re:Show HN: DeepSpeech based automated transcription service
by braindead_in on Friday June 29, 2018 @02:23AM
(#17422771)
The WER on LibriSpeech clean test set is around 0.087 and the CER is 0.030. We trained on around a 5000 hours dataset which included LibriSpeech train.
[ Parent
| on HN ]
Re:Show HN: DeepSpeech based automated transcription service
by gok on Friday June 29, 2018 @02:45AM
(#17422854)
So where did 95% come from?
Re:Show HN: DeepSpeech based automated transcription service
by braindead_in on Friday June 29, 2018 @03:00AM
(#17422906)
I think I made a mistake there. It should have been 92% accuracy. Can't find the edit button now.
Re:Show HN: DeepSpeech based automated transcription service
by gok on Friday June 29, 2018 @10:51AM
(#17424841)
Have you considered using a more traditional HMM style recognizer? The stock Kaldi chain model recipe should get you more like 5 or 4.5% WER on LibriSpeech.
Re:Show HN: DeepSpeech based automated transcription service
by braindead_in on Monday July 02, 2018 @03:50AM
(#17439945)
We actually started our experiments with Kaldi and even built a dataset out of our files to train on. But we found that Kaldi required a lot of data-prep and a long lead time. Our internal dataset is quite large and data prep is quite easy compared to Kaldi.
Re:Show HN: DeepSpeech based automated transcription service
by chadmeister on Friday June 29, 2018 @08:29AM
(#17424089)
You mean you meant to type 91.3% but accidently typed 95%? Were you using your own transaction software?