Okay, herewith in the context of rough backing track, very much a work in progress. The ‘chorus’ probably won’t make it to the finish line and the ai will be changed too, but it gives you an idea.
It does sound more like a real human voice that has been treated rather than a synthetic voice trying to sound human.
You’re singing this an octave lower right? I wonder if it were sung in head voice an octave up how realistic it might sound. Have you played with swapping in a male voice in your own range?
Yes I have and a my voice translates well to a male. Because I am singing at the correct pitch. Unfortunately because I’m a baritone (and have a really good falsetto) I can’t get up another octave in my ‘head’ voice. If I could it may help with male to female conversion - it would eliminate sibilants turning noisy and the s to z syndrome. But for me that’s not possible. My falsetto doesn’t seem to translate to female either. I suppose I could just do a ‘Jungle’ and sing the song in falsetto with baritone underneath in parts…would it suit the song???
I think what all this demonstrates is that you need a half-decent singer in the first place to get the most out of an ai.
Apologies for the delay in responding, having asked you to post it in the first place! I would agree that it sounds more like a treated human voice than a synthetic one, and definitely better now I can hear it in context. Wasn’t sure about the pronunciation of ‘reason’ with the hard ‘s’ sound ( i.e. as opposed to ‘reazon’, as it were ) … was that intentional?
I know the ‘ears of the beholder’ principle has already been mentioned, and clearly you know what you’re wanting this to sound like; but from an outsider’s perspective, I wonder if you might be over-thinking it? If you’re looking for a voice that sounds super-expressive, then obviously a different approach might be appropriate, but for me this works pretty well as it stands. Just saying, and as always, with all due respect.
Also WADR, with regard to your observation about needing a half-decent singer in the first place to get the most out of A.I., I would only agree up to a point. People on here have been very kind about my voice, but as I’ve mentioned, I myself do not have a very high opinion of it at all in it’s raw state. After having tuned it ( admittedly some may say ‘over-tuned it’ ), yes, it’s just about listenable, and an adequate basis for the A.I. voice that I used to replace it; but by no stretch of the imagination would I consider myself to be a half-decent singer without the aid of computer technology to pick up the not inconsiderable slack.
Hey that was a compliment to your voice Firefox7. Much appreciate your and everyone’s comments on this interesting thread.
I think the only thing for me to add is that I see AI as another form of vocal processing.