kahne / fastwer

A PyPI package for fast word/character error rate (WER/CER) calculation
MIT License
68 stars 15 forks source link

How to obtain the number of correct symbols? #7

Open 97morningstar opened 3 years ago

97morningstar commented 3 years ago

I would like to normalize the output following the article by Rafael C. Carrasco:

https://sites.google.com/site/textdigitisation/qualitymeasures/computingerrorrates

Like this:

image where C = Number of correct characters

How can I calculate the number of correct characters?

nabilafathia commented 3 years ago

I have this issue as well and can't find the solution anywhere

365id-github-admin commented 2 years ago

Any more information in this matter?

eslambakr commented 1 year ago

I used the following lib instead: https://pypi.org/project/cer/1.0.1/