MERaLiON

/

MERaLiON-SpeechEncoder-2

like 10

Follow

MERaLiON 108

Hi, thank you for the efforts in open sourcing the models
I am referring to the chart here https://github.com/openai/whisper?tab=readme-ov-file#available-models-and-languages
Whisper large has a WER of 18.3 for Tamil
But the one that is reported here is around 40.
The discrepancy seems to be huge. Can we get a clarification for this??

Thanks in advance.
Abhinav Ramesh Kashyap