In the paper, you mentioned that for MOS computation you "randomly selected 50 utterances from the LJSpeech dataset". Could you please specify which utterances exactly were used for MOS computation? That would be extremely helpful for my project devoted to automatic speech quality assessment.
Hi, thank you for the awesome paper!
In the paper, you mentioned that for MOS computation you "randomly selected 50 utterances from the LJSpeech dataset". Could you please specify which utterances exactly were used for MOS computation? That would be extremely helpful for my project devoted to automatic speech quality assessment.
Thank you in advance!