Word error rate of existing ASR APIs on Common Voice

[Franck_Dernoncourt]

FYI, here are the word error rates expressed in percentage for several ASR APIs on Common Voice (CV) v1 test set (cv-valid-test folder):

Note: it is possible that some ASRs have been trained on it, making the word error rates lower then they should be.

Code used to get those numbers: https://github.com/Franck-Dernoncourt/ASR_benchmark

[zeno (Zeno Gantner)]

Interesting. Have you also looked at Mozilla DeepSpeech?

[Franck_Dernoncourt]

Not yet, on the to-do list.

[Franck_Dernoncourt]

Interesting. Have you also looked at Mozilla DeepSpeech?

See Does anyone got a good result when training the Common Voice data set?