The part that talks about melismas is in 3.4.3, and the phonemes for English are in 5.3.
LOL. One thing you could conceivably do is use the formant control, possibly automating it if different parts of the song need to use different shifts due to vocal range issues. No clue how effective that would be, but I did find, in my initial experiment with it (using “The Star-Spangled Banner”, which is somewhat rangy, but not Mariah Carey rangy), I did find that I liked both voices better with the formant very slightly lowered, though I didn’t attempt to (or even need to) automate it over the course of the song. (I did automate various other performance parameters for expression purposes.)
This is a great angle on the topic. That would be really useful, if one could just enter sheet music, and it would song it right back to you, like a real vocalist would; that is gold man..
What is weird is - if you are in score and enter text in the text box at the upper right, that appears in key edit, but not under the notes in score. But they do appear in the same text box in score And if you put in words in key edit, the that same text box, they will appear in that same text box in score, but not under the notes.
So the workaround is to enter the words in that text box, then manually, using the v1. command, put them under the notes in score. Clunky, but it gets Minnie to sing the words..
Yeah, if Steinberg can get a link to working between the text box and entry via the V1. command, we’ll be flying..
Thanks for the link to the manual… I’ll work on it this week. If I can get it to pronounce French to any degree, I’ll let y’all know..
A Steinberg person clarified the non-relationship between Score Editor lyrics and Omnivocal lyrics (which go in a text field, not a lyric field):
My interpretation of this comment, coupled with your observations, is that the text field is just like a text instruction in the Score Editor, not lyrics, though perhaps the horizontal positioning of that text may be close to how it would be if it were a lyric?
To be honest, I’ve never used Cubase’s score editor for lyrics – I only use it as a visual aid for editing MIDI, with the score editor on one screen and key editor (for the greater control on editing) on the other.
I do use Dorico for making lead sheets, but, even if I’ve got a melody track in Cubase while working out an arrangement, I don’t bother importing that into Dorico as I want more precision in entering the lead sheets the way I want them, and what I’ve played will be more feel-oriented than printed notation precision-oriented. If I want a lead sheet for tracking vocals at all (I’m doing my own vocals on my own songs, and I’ve often played them live a fair bit before recording the songs so just need lyrics), I’d also have been using the lead sheet prior to that point for refining the melody and phrasing on the songwriting front. (I’d never done that with Finale as Finale was too much of a pain to use when making adjustments along the way, so I only made lead sheets after a song was completely finished, and usually that was after having recorded it.)
Now I understand why she wasn’t saying the words right – I was using hyphens for multi syllable words.. Much thanks for the manual..
Hyphens work in C13 lyrics, but they confuse Minny in Omni. Also periods (.). I’m sorting out pronunciation of “wasn’t”… I have it as 2 eight notes, and Omni will say “was” but can’t pronounce the “n’t” with or without the apostrophe.
Somewhat success. I used Fr. Phonetic spelling, and adapted that to Text input, plus adjusting the note length, to get her to sound pretty good. More dynamics, note placement (ahead / behind the beat) and phrasing, helps a lot. It seems like Omni doesn’t respond to pitch bend.. I wanted to get her to slide into a few of the notes. It would be nice to get aftertouch to control tone, too.
I think it’s cheating. But lots of what we do is also cheating.
As soon as we use someone else’s sample, or even modify our own vocal with effects we are technically cheating.
But what is cheating in this context? It’s posting a final product that is not what we are advertising it to be. If we say, “listen to my song” but it’s not actually a direct recording of the works of your own hands, it’s not really “my song”. It’s an amalgamation of multiple peoples’ work + multiple technologies; which were developed by computer programmers.
But the lines are so blurred that it doesn’t really matter at this point. If the market doesn’t care and if the law of the land doesn’t care then why add moral objections which aren’t there already?
…
When I use a content search feature to find something in my pdf book, instead of manually going through the analog paperback manually, I’m technically “cheating”. I did the work more efficiently with a tool, while someone in the past had to do it manually.
When I remix someone else’s sample that I didn’t record, I’m technically “cheating”. I didn’t record that sound but I’m benefitting from its existence.
If we were to get upset over every little aspect of cheating in content creation then we wouldn’t be able to do much at all anymore. Borrowing from other peoples’ work is so dug into our modern tech that it has to be acknowledged. Even in our DAW about screen it shows all the technologies that it uses that the company didn’t develop themselves.
So it’s the full-on SUNO trash that is the real offender; and I wish everything made or modified with AI needs the disclosure at point of sale/stream, enforced by law. That’s all we need. We don’t need to bring down every pirate and every snake in order to level the playing field. We just need them to not hide how they made it. Then the market can make their own mind about supporting that content creator or not.
In order for Omnivocal to sort of edge some of the existing competition like Synth V and its big brother Vocaloid, we need the full phonetic alphabet. Even with the supposed support for some languages, some words of those languages are not possible to be made because the phonemes necessary to make them are not there. We need full IPA support.