Open sh-lee-prml opened 4 months ago
As one of the author of UTMOSv2 the key difference of UTMOSv2 is a spectrogram feature extraction. Here, we used EfficientNetV2 as spectrogram feature extractor. Let me know if you need any help in implementing UTMOSv2.
I am also paying attention to the VoiceMOS Challenge 2024👀
From the description of UTMOSv2 repository's README.md
, UTMOSv2 seems to be good!
Soon after the publication of preprint papers (UTMOSv2 & Challenge24), I will check it and consider UTMOSv2 re-implementation in this SpeechMOS👍️
@Wataru-Nakata
Thanks for your kind help, that's reassuring!
Hi @tarepan
Thanks for your simple implementation again.
Today, UTMOSv2 was released. https://github.com/sarulab-speech/UTMOSv2
Could you integrate UTMOSv2 with HuggingFace Hub like SpeechMOS?