AI voices: Opinions?

What do you guys think of the AI voices so far? I love how realistic they sound, specially Solaria and Kevin. People who are not familiar with this software think it’s a real person and I love it. My only problem is how limited their range tends to be. They can’t go too high, or too low. So sometimes it’s hard to find a good song for them, because most of them go outside of their range. Also, the tension parameter only works if you want to make their voices sound whispery. If you want to make them sound super strong, you’re out of luck, because even if you turn the tension all the way up, little change will happen. I hope that’s something that can be improved some time. Nevertheless, I am loving doing covers with them, and listening to other people’s covers.


The AI voicebanks have been a real game-changer, but when Solaria came out, that truly shocked me. That took vocal synth to a whole other level.

How high are you trying to go? Emma Rowley’s vocal range is (not sure how accurate this is) is listed going up to G5. I don’t think I’ve ever gone beyond C5 in a song. I mean, who sings that high?

Personally, I’m not that impressed with Kevin. I probably would have been before Solaria came out, but now…not really. To me, he just sounds like some guy who you work with at the office. A guy who can sing, but he’s not opening on Broadway any time soon. I guess I’m spoiled now because I want all future voicebanks to be modeled after professional singers. Kevin doesn’t sound like a professional singer to me. I don’t know if too many professional singers would want to do that, though. I’m just super grateful Emma Rowley did this.


I think we should treat AI singers as real singers. Be aware of their vocal range and possibilities when
selecting songs for them. Or transpose the song.

I like AI voices so much that I don’t have any non AI VB (except lite ones).


Solaria’s voice supposedly can go up to a G5, but I’m using her now in a song that requires her to sing an F5 in full voice, and she sounds weak. Actually, anything after a D5 sounds weak and airy. I still like her tho, because her tone and how realistic she sounds makes up for it. I just wish there was a way to adjust the intensity of the note so you could add a little bit more of power to it. I do like Kevin, he sounds nice…but I remember they said his voice was fit for rock music, and it’s so not.

Right. The AI voices have the same limitations that a real singer has. I think I’m just being unrealistic by expecting voicebanks to have unlimited range…:crazy_face:

I feel like the voices are amazing… just that the current vocoder and engine has problems handling lower notes. If you go below mid c3, the voices start to get super grainy, even when they’re male and deep.

1 Like

Solaria has set a new standard for other libraries to live up to, I feel. I’m really looking forward to Xingchen Infinity, because what the two have in common is that they’re voiced by actual professional singers.

Kevin sounds good, but he reminds me a lot of UTAU libraries that have been recorded by someone speaking rather than singing. The overall quality of the library goes up when the voice is provided by a professional singer.

Nekomura Iroha was leaps and bounds ahead of other VOCALOID2 libraries, and that’s simply because her voice provider was a professional singer, with a crazy good control over his voice. AHS subsequently didn’t mess too much with the samples, and Iroha, while quirky, has a fantastic vocal quality.

Going forward I’d love to see more singers providing vocals for libraries.

1 Like

Agreed that Solaria is a game changer. I’d love to see a great, deep English-singing male voice. I have Kevin… but his useable range isn’t much lower than Solaria.

I’d love to have a way (or another version of a library) to overdrive the voice. None of the tuning params or scripts have been able to achieve that IMHO. I expect it would simply require another take of every syllable from the singer and maybe a tuning transition between the two. I’m willing to pay for that :smiley:

1 Like

I ADORE THE AI VOICEBANKS, i especially love the dreamtonics voices god theyre so good

1 Like

Yeah Kevin doesn’t do it for me. You can clearly hear the difference in library quality when the voice has been provided by a professional singer (Solaria, Xingchen, Weina) and when it hasn’t. The libraries can still sound fantastic but they don’t have that quality about them. With vocal modes and HDVM Solaria is just immensely capable.

I’d love to see a voice provided by a really talented male singer, like Colm McGuinness, Jared Halley, or Chase Holfelder.

This post was flagged by the community and is temporarily hidden.