Vocal recordings from XML! Cantamus

Has anyone seen this, or tried it with Dorico exports??

Here’s a quick demo. I uploaded a basic XML to Cantemus: literally just simple notes and lyrics, no tweaking. This is the result I got.

I think this is pretty remarkable, considering what it is. And considering that I did absolutely no work on the file beyond inputting notes and lyrics.


Never heard of this before. I’m blown away by the fact that the rendition is intelligible and palatable. Makes all the effort I go to on my channel seem almost hard to justify.

I’m currently uploading a five-part acappella arrangement of mine to test it. Shoot, it’s really, really good.


When they’ll have a VST, it’s goodbye to Wordbuilder and such…


Damn! What do they use to achieve this? :astonished:

1 Like

There’s obviously some machine learning going on, and they mention creating models of real singers’ voices (sounds like Matthew Curtis is one; he’s not my favorite, but he’s everywhere, so I can’t say I’m surprised).

Well of course! It would seem a very evolved form of the East West Symphonic Choirs … interesting! It would be great to have it as a VST that you can use in Dorico … :thinking: :slightly_smiling_face:

1 Like

I suspect that to create a VST they would need to host it standalone and employ a delay just like Noteperformer.

Emvoice One is another means of adding sung words to MIDI. It can work in Dorico but is a more comfortable fit with Cubase, since Dorico cannot yet send tempo information back to its VST’s. Thus Emvoice piano roll does not line up with Dorico bar lines.

Let’s say that more than anything else it is a hope for the future :grin: … but apparently the technology runs fast and maybe … :man_shrugging:

1 Like

The technology runs fast, but the challenge is considerable and has a long way to go. That is already true of simulating live playing of instruments, and adding pronunciation and appropriate emphasis of lyrics is only more varied and thus difficult. I am just glad to see several companies working on this and am impressed with the progress being made on such a difficult challenge.


Of course … I am well aware of the difficulties … for now I would “settle” for an update of the Symphonic Choirs East West, which at the moment if I’m not mistaken is the only vst that can handle words in a satisfactory way … .

OF the players that offer lyrics for choral sounds, EW seems the one I have heard the most about. Have not tried it yet.

The thing is, all those take a lot of work. This is literally just plug and play. Absolutely zero extra fiddling.


Just tried this out and it is quite amazing. I’m very impressed…miles better than the Emvoice One renditions that I have heard. Certainly it’s not a choir sound, rather single voices, but I will experiment with rendering the individual voices, capturing them and trying to layer them to create more of a multiple voice per part sound. They really need to create a plugin! A DAW will be the only way to assemble all the parts, but you can absolutely understand the words. The EW choirs are indeed quite fiddly to get a good result, but it is currently the best thing we have for choir renditions. I also have Dominus Choir Pro from Fluffy Audio which is a beautiful sound, but has limited word creation. Good for Latin but not so much for English. Cantimus is certainly very encouraging for the future.


Of course, because it’s not meant for choir renditions.
It’s meant for individual practice, as it says.

A plug-in is not feasible, though. Rendering can take as long as 30 minutes for one piece. Not sure how they would adapt the engine for real-time playback.

1 Like

Yes, I’m realizing that as I look into this further. As far as creating a choir is concerned, I rendered the same small vocal line four times, using a different voice for each pass. Even the bass voice singing a soprano line is useable it seems. I put them into Cubase and offset the tracks slightly, even though there was no initial phasing - each rendering is different. The result is pretty good. Using this method I can easily create an SATB choir with four voices on each part. The problem is the number of credit’s available using their subscription model - particularly as this is not how the service was intended to be used, as Mark pointed out! I’m sure that if the subscription model was changed to allow the purchase of extra credits as they were needed, this would be a very successful venture. I have written to them to ask for a different payment model that composer’s could use. I’m happy to pay for credits in order to render my own music, but right now there is a yearly limit as I understand it, and I think I will quickly need more for the chamber oratorio that I am currently writing.

You don’t have to have a bass voice sing the soprano line! You can select which voice is assigned to each staff at import.