Originally Posted by Guitarhacker
So the vocal thing..... That wasn't a Suno vocal. It was from Audimee.
That's why I also mentioned Vocoflex. All these vocal cloners without UIs will run into this problem, and there's no good way to correct them. frown

Quote
I used Suno to grab the cover of the song and used the drum track from Suno but tossed the Suno vocal in favor of Audimee. There's always going to be issues with understanding and what it actually renders.
Sorry, I mis-read your description as a Suno vocal then loaded into Audiomee. Which, of course, would simple increase the chances for errors. blush

Quote
As a side bar.... the song I did previous to this, one of the words in there was wonky. I rendered it several times and each time it came out weird. So I did what I needed to do. I colored outside of the lines. If it won't give me what I need, I will make what I need. So I searched the song for the "letter/sound" I needed to start the word correctly, I trimmed off the start of the wonky word and inserted the "letter/sound" that started the word, pitch corrected it to the pitch of the word and amazingly, it sounded perfect and no one listening, made any mention of that word.
Heh. I've done that with my own vocals when I was too lazy to re-record them.

I think the word you're looking for instead of "letter/sound" is "phoneme".

Quote
Nothing in this caught my attention but that's the beauty of having others listen to the parts with a critical ear. Other folks hear things that I don't, or, simply ignore.
There's been more than a few times I've put typos into SynthesizerV. Or worse, sung a lyric wrong when doubling/harmonizing a line, and not noticed until much later.

Quote
I did hear a "L" sound in that point you mentioned. Who knows what happened.
Well, /l/ and /r/ are "liquid" consonants, along with /w/ and /y/. That is, there's no static target that identifies the sound, so it's the movement that makes it distinct. The phoneme recognizer picked what it thought was the best match.

Early speech recognizers weren't very good until they were augmented with word probabilities to guess at what the user was likely to be saying. But I'm pretty sure the phoneme identifier in Audiomee doesn't use an English dictionary.

-- David Cuny
My virtual singer development blog

Vocal control, you say. Never heard of it. Is that some kind of ProTools thing?