I am using Ace Studio AI to generate vocals. It’s a bit lacking in execution but it’s a new concept of late.
It would so nice to be able to have such a plugin for Cubase. Lay a track in midi and then type in the lyrics in the notes.And choose a singer (other singers could be bought separately).
Maybe it’s too expensive to develop , but it would be nice
Have it all in my favorite DAW.
Hi,
Add the optional feature-request tag to your post, please.
Exactly this:
I cant think of why you would want to generate a vocal. But maybe thats just me
2 Likes
We are rapidly heading in this direction…
https://m.youtube.com/watch?v=eKxNGFjyRv0&pp=ygUVYmVhdG8gdGhpcyBzb25nIHN1Y2tz
Because you might be writing a song and have no vocal skills to speak of - like me - and want to get an idea how it might sound.
1 Like
I wouldn’t think of this sort of feature request as a potential Cubase feature. It’s too specialized, and, beyond the underlying technology, there is the issue of licensed voices, which would seem to me to go well beyond what makes sense at a DAW level. And, of course, there are already a number of third-party products, including plugins like the above-mentioned one from Dreamtonics, specializing in this sort of thing.
What struck me a bit ironic on this topic, though, is that Steinberg’s parent company, Yamaha, was an early pioneer in the sort of product being alluded to (though I don’t know AI was involved at that point). I was curious to check what the current state of Vocaloid might be. Was it even still available and an active product? Yep:
In fact, their “how to” type information even has a instructions for setting it up in Cubase 12 or later:
My own exposure to Vocaloid came back in 2004 when I reviewed Vocaloid MIRIAM for the now-long-defunct CakewalkNet e-zine (print image archive of the review can be found at https://rickpaulmusic.com/wp-content/uploads/articles/zero-g-vocaloid-miriam-review.pdf), though I’d initially started working on a review for Vocaloid LOLA, a slightly earlier generation product.
I’m not going to suggest Vocaloid is in the ballpark on what I’ve gleaned of Synthesizer V – while my impression is Vocaloid has come a long way since I tried it, the demos on their site don’t sound in the same league as equivalents I’ve seen from Dreamtonics or results I’ve heard from people who’ve used Suno. Also, while I don’t know how natural (or not) it may be to use Synthesizer V as a plugin in Cubase, the setup guide I linked for Vocaloid in Cubase decidedly has a lot of hoops to jump through. (When I tried out Vocaloid, it worked in SONAR via Rewire, which was also not particularly fun, though somewhat similar, in concept at least, to how ARA2 works now, and that does seem to come into play with Vocaloid6.)
With respect to the question of uses, my initial hopes for Vocaloid were twofold:
One was when I wanted a female voice for a song demo. I did end up trying that out with Vocaloid, and it was okay for an early work demo in a key too high for me to sing, but not the sort of thing I’d be inclined to pitch to anyone. That sort of use could still be something I might be interested in certain context today, though it’s nothing I’ve even attempted to try to date.
The other was to get some different timbre voices to layer with my own for background vocals in order to get a richer sound – kind of like having multiple singers do harmonies instead of just the same singer who did the lead. I did actually use Vocaloid LOLA this way on one recording (“Undertow”), though mixed pretty far back, largely due to the artifacts of that early product. Since then, I’ve attempted lots of different ways to achieve similar ends based on processing my own voice (e.g. formant shifting, pitch shifting, some of the iZotope Nectar’s Backer voices, Waves Harmony, etc.) and blending processed vocals with more natural vocals. This is an area, though, where I do think this sort of feature could be useful and would likely provide better results, though it is again not something I’ve even attempted to try to date.
In my mind, while I don’t think I’d be likely to want to use an AI vocal for a lead vocal for a recording I’d put out there (other than as a song demo geared toward promoting the song for cuts by other singers), I view the potential for using it in a backing vocals context similar to using virtual instruments for real life instruments that I don’t actually play, which I do all the time. (I’m mainly a keyboard player, so, for example, I make heavy use of virtual guitars, some of which I play directly on a keyboard and others of which do chord recognition and play loop-based samples based on the chords I play or program.) But it would heavily depend on having access to a variety of voices, all licensed for potential commercial use, and with reasonable costs and not overly complex user interfaces.
1 Like
This exactly. Well written…
1 Like
I am a home studio musician. I don’t sing.. i don’t know any singers.
I have been writing instrumentals in cubase since cubase 4.
I just did 2 songs, which i feel would add another dimension for my hobby.
that’s why
Ace Studio is a cheaper version of Dreamtronics, sort of like what Photoshop Elements is to Photoshop.
It does a pretty good job, although you have to monkey around sometimes.. like when consonants are not heard, or the word sounds funny.
They have just released a BRIDGE which works as a plugin to cubase. I haven’t tried it yet.
basically, i write the lyrical melody in midi.
I first do an audio mix down export of the whole song.
Then i export the midi (of the vocal track).
I import the 2 tracks into ACE … do the synthesis and then export the resulting vocal track back into cubase.
I would post my first song, but it was my first and there are missing consonants (or not clear ones). I may re-do the vocals, since i have improved my ACE vocal track on my current song (The old folk song, Scarborough fair re-mixed for EDM’ish)
lol. How a Justin Beiber vocalist
He is right. But I’ve heard a lot of really crappy songs with crappy lyrics.
If you write stuff for the masses, there’s nothing wrong with churning out AI generated music. Just like my old hobby stock photography… I had to place my 88 year old mother on the floor (help, I can’t get up)… now I could just ask AI to generate it (well this is in its infancy so it will get better)…. And people will buy the photo and not care if it is generated or not.
TL;DR
The Vocaloid 6 setup for use with Cubase is straightforward; the vocal synths are ok (they’re not real vocals!); the editing process with Cubase is akin to MIDI editing.
Detail
The integration of Vocaloid 6 (V6) and Cubase is straightforward, despite what the docs say! I’ve used V6 with C12 through C14. The original setup was simple, it’s survived the Cubase upgrades, and has been stable in use.
V6 has a base set of around 18 different “vocalists” (you can purchase more), each with slightly different sonic qualities. The sounds are comparable to Dreamtronics Synth V (imho) – the V6 demo songs are a bewildering collection, in that they don’t showcase the capabilities very well.
The vocal synths sound better if you put the effort in: judicious EQ, de-essing, triple tracking, well-crafted harmonies, captivating lyrics etc. .
Vocaloid 6 can be used in its own editor – sync’d to Cubase via a bridge that’s largely transparent in use – or directly as an instrument in the DAW using the “VX Beta” plugin. [VX Beta is free … if you purchase V6]. I believe Synth V works in a similar way to the V6 editor. Using VX Beta as an instrument is a bit clunky (it’s a Beta product!). Both ways (bespoke editor & integrated instrument) allow for lyrics and melody to be input – the bespoke editor is a bit slicker when tweaking the vocal sound.
n.b. this process – you create the lyric and melody for the vocal synth to “play” – isn’t the same process that products like Suno use: which creates the melody (and everything else if I’ve understood it correctly) from provided prompts.
Personally, my current preference is to create songs (largely synth-pop style) using V6. I occasionally sing myself but mainly choose to use the vocal synths: I like their formant glitchiness. The vocal synth sound isn’t to everyone’s liking – my partner says they sound like “wailing banshees” 
. [Examples on YT via my profile, if you’re curious.]
Apols for rambling on, happy to clarify if needed.
2 Likes
i appreciate your rambling.. It is so rewarding to compare notes with all the Cubasists here in the dungeon.
I use a plug-in, called EmvoiceOne, which is great because it doesn’t need any kind of bridge and it understands variable Tempo, unlike other vocal synthesisers, which if they did (allow variable tempo) I would sign up, simply to obtain more voices but alas they don’t unfortunately.
Thanks for the current notes on Vocaloid 6. Interesting. I did check out a bit of your YouTube (bits and pieces of the first three songs on your channel) to get a sense. I definitely don’t hear the “wailing banshees” comparison. The voices seem pleasant, and they don’t have the strong accent that early Vocaloid singers had.
In general, it seems like a massive improvement over where Vocaloid was back when I was checking it out, both “back in the day” and occasionally at times over the years. (If you want to get a laugh at some of the early Vocaloid stuff, search for “make me feel” along with “vocaloid”, for example on YouTube. That will lead to a number of Vocaloid versions of one of my songs that apparently were popular enough among the Vocaloid community to merit a page on the vocaloid.fandom.com site. Some of those were done by my collaborator on the song, Alexei Ustinov, who I know did at least one version with Big Al and Sweet Ann, but others I’d never heard of have done covers, I think mostly of just the same clip that Alexei had made with various Vocaloids in different musical styles. I eventually recorded a live singers version with Beverly Bremers, with Alexei co-producing the tracks with me.)
That said, I also noted some of the same sort of glitchy artifacts (akin to unintended grace notes) that Vocaloid always had in the early days, though they were much more frequent then back then versus only hearing a few cases here. It also felt like the emotional level of the “singing” on all three songs felt pretty flat. Mind you, it may be that’s what you intended for the specific style of music, and it may be that, with more work, that can be addressed. (The thought caused me to take a closer look at some of Dreamtonics’ tutorial videos, and their stuff also starts out very flat emotionally, but some later videos showed ways to manipulate this, though it felt like it would probably take a good deal of work.) Perhaps one day I’ll get curious enough to try the V6 demo version, though probably more just for curiosity than likelihood of purchasing (way too budget-strapped at this point to even consider that).
And so begins the death of real music.
Well, it’s a contentious issue. I am against AI, but I did use an AI vocal, can’t remember the company, to demo a song to a film director.
Once it was approved, we used a real singer, which BTW was 99.9999% better than the AI vocal.
If one wants to demo a song, I can’t really see the harm in that. Already there are things like chord assistant and sampled loops etc, which is stuff that I am not that enthusiastic about.
However, if one plans to release a song commercially, then it’s when the issue starts,
For the love of God, get a singer. There are websites out there such as AirGigs that offer vocalists at great price.
In addition, that AI vocal that one will release commercially, will be owned by the AI company, so “your song” , will not be “your song” as the AI company will claim a cut in its discretion.
So…it’s all fun and games, but far from clear cut , or simple.
The good people who make the EmvoiceOnce plugin charge nothing other than the subscription price to use the plug-in, there is nothing else to pay at an any stage when a song is released that features any of the voices available using the plug-in.
Terms and conditions change all the time, and Companies don’t give you things out of the kindness of their hearts.
I wouldn’t touch an AI company for something like that with a stick…
1 Like