Talk:UTAU/@comment-27324770-20160522022616/@comment-29714788-20160522111409

Also it really depends on the resampler you're using with the voicebank, I find that moresampler works well especially for power voice types and produces a clean sound, it has a bunch of flags which I haven't used a lot but are pretty cool. EFB-GT and fresamp are also good for power voice types, fresamp is good for soft voices too but I personally don't like moresampler for soft voice types because of its engine noise which actually reminds me a lot of Vocaloid's. TIPS works best for low voices but it sounds metallic, probably because it's a stretching resampler and not a looping resampler like tn_fnds. For soft voice types I prefer tn_fnds over anything, but compared to the other resamplers Vocaloid is better at handling them which is why I like soft voice types in Vocaloid more. But Vocaloid makes power voice types sound either nasal or metallic, like Ruby or Flower, or both, like Cul, so that's why Yukari Lin is my least favourite out of Yukari's V4 voicebanks. I don't know too much about bkh01 or phavoco because they're kind of weird, and w4u never works for me (apparently I need to compile fftw files for it which I'm not sure how to do). That being said, I think a lot of people exaggerate UTAU engine noise because it's definitely more HQ sounding than Cevio or Alter/Ego imo. It's not possible to compare it to Cevio in the first place though, because UTAU is concatenative and Cevio is HMM, I'm not sure which synthesis method Alter/Ego uses. By the way I'm speaking from experience with low/medium to high quality voicebanks, of course very low quality voicebanks will sound horrible no matter the tuning.