Hi all,
The new Omni vocal plugin is a great taste of things to come!
I’ve been mucking around with it, and although it’s in beta, I’m unable to get a male voice out of it. No matter which one I choose, it’s always female. Are others getting the same result or have found a way around it?
Thanks
Hi,
maybe try a lower register? This might as well be a matter of the octave the voice is in and your expectation on a male sounding voice.
Hi Reco29
Thanks for that - I will give it a go.
I have the same problem. Tried everything. Reinstall, different registers etc. It’s a bug I suggest and will have to wait for Steinberg and Yamaha to fix.
It works fine for me as long as I put it in the proper octave.
The transpose and fine tune dials enable you to set all the equivalent human voice ranges from soprano through to bass. Experiment until you find the pitch you are after. It helps if the original music that Omnivocal is following, regardless of whether being written in the piano roll or score editor, is also of the correct register for the pitch you are after, if this proves to be the problem try lowering the notes accordingly. The other dials in Omnivocal allow you to vary other aspects of the voice such as tone, power etc.
I find Omnivocal quite good, considering it’s in Beta, once you get used to it. One thing I have learned is that the dials are sensitive, voice is an instrument just like any other and it can take a while at first to reach the exact sound you are after.
I hope this helps.
No problem here, too. Works pretty well for a beta version 
Did you ever get the male voice to work?
I’ve not.
I can’t any voice out of it. WTF how come I see so many vids of this working?
I can change voices and it works but once I have entered any data (midi) it doesn’t work. The picture changes but the voice doesn’t.
Windows 11
Hi Rick,
Not really - I followed suggestions to sing lower notes and that sort of does it, but singing the same notes, both male and female sound identical.
There seems to be limits on the ranges of both singers beyond which they don’t sound natural for their nominal genders. I just tried an experiment a few days ago where I’d done demos with both the male and female versions using the same song and the exact same automation on Omnivocal, but put the singers in different ranges – key of C major for the male and G major (up 7 semitones, specifically) for the female. Just now, I tried another experiment where I went back to the version where I’d used the female voice and switched it to male (no other changes). There was an obvious difference in tone/gender in ranges that were common between the two voices. However, once the male got above roughly the G above middle C (which had been the highest note in the male version of the song, the lowest being the C below middle C), the male started to sound unnatural and more feminine. I didn’t try the reverse experiment with the female voice in the male range, but I do remember when experimenting with it earlier that, when it got to a certain low range (I don’t recall where the break was), it started to sound more masculine.
FWIW, if you’d like to check out both the demos I made with the two Omnivocal singers, and my notes on the process I used for doing that, here is the post:
One thing worth noting that I didn’t think to mention in my earlier notes: I found that the male singer sounded more natural if I turned the formant down to 46% instead of the default 50%. I also used the same formant setting on the female version (I didn’t even think to try otherwise when I was just quickly trying to make a female version for comparison).
I tried every suggestion I could find and I get low notes and high notes, in every shape and form, but it’s all clearly female…
Also, the suggestion to keep male voices on low notes to make difference in sound is never going to work. A decent metal singer can reach an E5, and there are even male singers that reach into the 6th octave. And with a duet with a female and a male voice, using the same register, you’re gonna need different sounding voices of course.
The last post is like 6 months ago, with no more complaints. Does this mean there is a fix by now?
No, no changes in this regard.
I can imagine two possible future plans for OmniVocal. Either it is just a gimmick that was added to Cubendo because Yamaha had it lying around or, which seems more likely to me, Yamaha/Steinberg are currently working on getting the base product in good condition before Yamaha/Steinberg will release additional voice banks for additional purchase.
The voices in Omnivocal aren’t exactly “metal singer” types. 
The thing is, with any voice modeling, be it for vocal synthesis or for voice replacement (e.g. IK ReSing), each model is going to be based on some specific singer, with some specific (and/or context-sensitive) range. (What I mean by context-sensitive is most singers have a chest voice range, a head voice range, some sort of falsetto, a maybe lower range when they have a cold, etc., and the voice characteristics in each range can vary. I think of my falsetto as my “Mr. Moose” voice, referencing a puppet in the old Captain Kangaroo TV show that was popular for kids when I was growing up. That is decidedly NOT a flattering comment on my falsetto, which really isn’t very good.)
My impression of the models in Omnivocal is that they are intended for a young pop sound, and, based on the experiments I mentioned above, my sense is both singers have a limited range – if you got above the G above middle C the male singer started sounding more feminine, for example.
In my opinion, there is no fix because it was never broken. Rather, there are limitations, both with only having two voice models and in the specific characteristics of those models and how they might, or might not, be useful for any given musical genre or specific composition.
Aside from my one “Star-Spangled Banner” experiment, just to make an early test of what Omnivocal could do (and how easy, or not, it was to use it), I can say with a high degree of certainty that I’ll never use Omnivocal for a lead vocal in one of my recordings. While I did use it to layer with my own background vocals on my most recent single, it was a lot of work using it for that due to the need to create the MIDI data to match my BGVs sufficiently, so I’m not likely to use it in that way again, either. (In the two recordings I’ve worked on since that single, I’ve used ReSing, which can just use copies of my background vocals directly and change the voices to add the different textures).
What I do still use Omnivocal for is placeholder lead vocals (and sometimes background vocals) at early stages in my projects. Being able to have lyrics at those early stages, and a somewhat more singer-like tone, is a lot more helpful than picking some vocal “ooh” patch to use for the same purpose while working on an arrangement, and I can also use it during the songwriting process if, for example, I need to demonstrate lyrics I’m writing to a collaborator who wrote the melody. And I don’t like to track my own lead vocals until I’m pretty sure I’ve decided on a final tempo and at least have the feel I want for the arrangement.
There are no two voice models, that’s the problem. There’s only the female voice.
And I dont need anything fancy, indeed just a placeholder. It’s just meant to create an example melody to hear while composing. I just need to be able to hear the melody with a voice sound, so I can recognize, in a full composition, which melody is which part. Anything so I dont have to listen to the hair-raising, shivers down my spine tenor from Voices of Rapture. I dont need power, I dont need a metal sound, I dont need lyrics, just a melody that is recognizable as either a male or a female voice.
I had this issue. If I changed gender during a session it would not change.
If I started a new session it would change.
I haven’t actually tried it for a while now, and not since the last update. So don’t know where it is now.
I’ll have a play and report back tomorrow.
Definitely not true in my experience. I have mostly used the male voice for my placeholders (since it will eventually be my vocal). FWIW, here’s a link to the early test I did with it with the male voice:
If you’re not getting the male voice, you may want to try reinstalling the software to make sure both voices are getting installed. From a quick Google, I find the voice banks are installed in:
C:\ProgramData\Yamaha\OmnivocalBeta\voicebanks
There are three subdirectories there with the actual data. What I’m seeing in mine are 2 files in the image subdirectory, three in the definition subdirectory (two of which have the “.vdef” extension, and the other one being default.txt), and 6 files in the model directory (two of those being just under 25 MB and thus likely the main data for each model)
One other thought: I found that with both voices, it helps to shift the formant down very slightly. But that’s not a gender-bending thing, just a slight difference.
I’ll put on my tutu.
