The voices just sound less fake than the other TTS tools I've tried, including ElevenLabs. I can still tell if I listen for it, but it's close enough that I stopped re-recording every line. I use it for short explainer and demo voiceovers. The script changes a lot, so I was spending more time on the mic than on the actual video. Cloning from a short clip is pretty convenient too, the voice stays consistent across videos. I mess with the emotion tags when something comes out too flat, though those don't always land on the first try. Review collected by and hosted on G2.com.
I end up regenerating a lot more than I thought I would. A line sounds fine except one word comes out weird, then the next take messes up a different word. When I'm trying to finish a video the same night, that gets old fast. I still mess with the emotion tags but I'm kind of guessing half the time. I wish I could keep the parts that already sound good and only reroll the bad word. Review collected by and hosted on G2.com.