>>47717680>V2 during Vedal's 3 minute test https://youtu.be/xZTn2IgVpgs?t=9https://youtu.be/xZTn2IgVpgs?t=59https://youtu.be/xZTn2IgVpgs?t=152>V2? during Anny's collabhttps://www.twitch.tv/videos/1797863356?t=03h02m28shttps://www.twitch.tv/videos/1797863356?t=03h03m07shttps://www.twitch.tv/videos/1797863356?t=03h05m05sDoes she sound more monotone and less emotional on Anny's stream than before?
If so that's a huge step backwards, because her AI voice's emotive potential and rich inflection is what made it so good compared to Azure TTS to begin with.
If the people saying V2 "needs more work" do so because they subconsciously want it to sound like a more primitive software than that's disheartening.
If you wanna ignore the new voices own unique strengths just to make it sound like a TTS anyone can use, then you may as well save the effort and stick to using her original voice then.
A shame.
One of the main appeals for Neuro is her receiving upgrades, but how is Vedal going to deal with real meat stuff like updating her language model once new stuff goes public?
If more and more people like her current robotic self so much they'd rather want her to stay who she is than become more advanced, expressive, and humanlike, we have a genuine problem that'll only get worse as time goes.
If Anny had a part in any of this, this just feels really bad, if hypocritical at the same time.