utmosv2

[sametcagan]

I am not able to use utmosv2 as we use pseudo_mos. I checked UtmosV2 repo for "from utmosv2.dataset.multi_spec import process_audio_only_versa" and I couldn't find process_audio_only_versa.

[ftshijt]

Thanks for your interest in using the toolkit. For the problem, may I kindly request you to check if you have used the installation scripts at https://github.com/shinjiwlab/versa/blob/main/tools/install_utmosv2.sh to install utmosv2?

Noted that we use our own fork to process utmosv2, which has the customized function for versa interface (i.e., process_audio_only_versa). For more information, please check https://github.com/ftshijt/UTMOSv2/blob/main/utmosv2/dataset/multi_spec.py#L24-L49

[ftshijt]

For a bit more context, using our fork, we can also make sure that the score is consistent by avoiding issues like sarulab-speech/UTMOSv2#40