openspeech
openspeech copied to clipboard
Testing performance of the re-produced model
Thanks for your great codes! Can you provide the re-produced results such as Conformer, Transducer? I suppose this is important for users to measure the quality of the codes.
I tested conformer-with-ctc architecture.
This architecture record 2.66 CER in LibriSpeech dataset. (Greedy Search, No Language Model, No Data Augmentation)
Thank you.
It looks like pretty good results! How about the WER (word error rate) at the test-clean dataset?
I'm afraid I've heard from one of the contributors, so I don't have any more information.