Today was the first time I put my hands on Omnivocal in Cubase 15.0.20 (OV 1.2.5 >> STILL BETA!) and I don’t get it.
I RTFM and created as many notes as the words in the lyrics were supposed to be “sung”.
I have a 8/8 / 55 bpm acoustic guitar piece and the lyrics follow mostly 4 beats.
Text as simple: “I am so lonely and want to be heard”
What happens > First a very direct, loud “I” then followed by an almost whispered “am” and a “lonely” which is something in between and strangely stretched. So I gave the notes a few adjustments to try to get it better and it all ended up in a mess. At one point it almost followed the notes but “am” was silent as the lambs, the last word “heard” was loud shout and in the middle it whas this and that. After almost 1 hour I made my mind and could only find out that it is useless. And, just to make it clear, with no text, so that only i.e. “Ah” as default is used, the differences in loudness was the same as with text, I can only guess that the weak algorythm is trying something to create it shouldn’t. And the value of the velocitiy does not change a thing and fiddling with Expression, Format, Attack etc. does not make things better.
I didn’t want to give it up for now so I added as insert FX the Cubase Vocal Chain and chose a preset where I removed just the delay/reverb ‘n stuff (I like the Vocal Chain Plugin) and yes, it made it better, as if you compress a poorly recorded vocal. I’d like to have more concrol of the dynamics though already from the plugin. Maybe one day it will be better if it isn’t BETA anymore?
I haven’t tried entering any text with the latest version (from Cubase 15.0.20), but I do think word volume is sensitive to MIDI note velocity. Could that be making a difference? (I am using a part I’d set up in an earlier version of Omnivocal with the latest version, but I can’t listen to both in parallel to see if there is a difference.)
With respect to the “strangely stretched”, I have also seen that, depending on what phonemes are involved in a given word, length of the MIDI notes the word is attached to can be important. For example, if the note is quite short, Omnivocal may not fit the a phoneme in properly, but, if you extend the note a bit, it will be okay.
FWIW, I mostly use Omnivocal for temporary lead vocals while working on an arrangement, so I am not super fussy on its pronunciation (which can leave a lot to be desired). I’ve also used it to blend with my own background vocals in at least one recording to date, but that was a lot of work to get the MIDI to match my vocal (just extracting MIDI from the vocals was more or less a mess needing a LOT of cleanup). But, even there, I haven’t been super fussy about pronunciation due to the fit into the mix. I’ve used Waves Sync Vx to tighten rendered Omnivocal parts to my own parts, and I do process the Omnivocal parts similarly to how I process my own BGVs in that context.
1 Like
Thanks for sharing your experience with OV. Currently I am planning to give it another try with my song, I may share it if it is of any interest. Might be that OV requires a bit more attention to get used to it. Although I still don’t like to have any BETA in my purchased prodcut (to be fair, they at least didn’t try to sell it as a RC 
) but that’s another discussion.
In my experience, it definitely requires some attention to get used to it. Also, it was mandatory (for me anyway) to download the user guide and learn about entering phonemes directly as it doesn’t always pronounce words the way I want them pronounced. (I’m trying to think of a concrete example, but I don’t recall for certain. One example might be “our” and making it into two syllables – like OW-uhr – and my needing to specify that it should be one – like the word “are” – but I don’t recall for certain. I just know that there were examples similar to that.)
Yeah, that did seem strange to me, too. I have no insight on that, though at least it is a fair bit easier to use than Yamaha’s Vocaloid was back when I reviewed it for CakewalkNet in 2004. And I have actually ended up using it on one finished recording (just doubling my background vocals), in addition to using it for placeholder vocals while working on writing and arranging on multiple productions. But maybe it will get released sometime between now and Cubase 16. 