I prompted an acting theatrical play conversation, with a cop and a witness who saw a bank robbery. I have prompted a Southern woman voice also.
I am just wondering if this can be used to create voices for Iclone characters.
It sounds more authentic and realistic than text to speech. But you might not get what you want. Riffusion is free with unlimited use. Here is a link to the audio.
Monologue female 1950 talking about prices
If you save the voice as MP3 or WAV file, you can use any voice with AccuLIPS. Just load the voice file and create a file with the text of the dialog (with the same name as the voice file and extension .txt) and it should work.
I use ElevenLabs voices that way but new options come out regularly.
EDIT: Just remembering: I think you use iClone 7. It would work similarly but the lipsynch is not as good. AccuLIPS is only in iClone 8.
Looks kinda cool. I can’t seem to get just “spoken/words” without the music. Any ideas?
Although I was not impressed with the quality of the voices
on the site that the op linked,
here is a site that uses AI to separate vocals from
their music tracks
Caveat: you have to have a paid account to actually download the separated tracks but there is a work around.
They let you “preview” your separated results so just use
any free screen recording software ,like OBS studio,
and capture a video (with audio) of the separated voice tracks playing in preview
and save out that audio with your video editor for use in Iclone or CTA.
That is how I made this old couple sing a duet in CTA
from an AI song I generated on UDIO.com
That’s cool! Thanks!
an additional note to separating vocals - the virtual DJ software - which has a free version has a technology that separates tracks, so you can literally create stems from any song.
it’s fun remixing classics that way - it separates drums, vocals, bass and keyboards as individual outputs - although you’d have to record it in real time because it’s a dj plug in for performance and not directly a plug in for breaking down tracks
the basic version is free but if you want to attach a physical controller to it, you have to buy the drivers per attachment - and if you want to remove any visual watermarks on the stream, there is an additional fee…but the extras are not necessary to do basic functions.
Downloaded and installed virtual DJ
its complex and deep but I figured out how to separate the voice from the music pretty fast.
a bit disappointing that it has no “export to .wav” function but I can just capture my separated tracks audio with OBS.
Thanks
it does - but it’s a live performance plug in - so you have to record the mix as if you were playing a song live, it’s very powerful though and does an amazing job at sifting through the tracks. it would be easiest to record the live stream though with a 3rd party software - I use an older version for audacity for that function myself. ( side note - don’t use the latest version of audacity- the last version 2.4.2 before it was sold to a new company is the best, the new company adds some spy / tracking features for capitalist reasons! 
)
Audacity can break music into 4 stems (Bass, Drum, Vocal, Everything Else) using the FREE Intel OpenVINO Music Separation plugin. It did a good job on drums, but in my experience the vocal separation was not the best.
Here is another monologue, unaccompanied female speaker. She is talking in front of an audience about the time she caught a foul ball in 1964, at a baseball game. No music. I think I am getting better at prompting for no music and only words. God Bless!
Original song created by suno.ai
Vocal track for lipsinc animation separated by Virtual DJ
