Does the Cantai playback template do anything except the space type? And isn’t there an expression map for it?
When applying the playback template nothing changes (as far as referring to an expression map or instrument, voice etc.). I would think there would be an expression map for this plugin considering it’s range of features. I haven’t been able to find any documentation yet, but I would think it also would be receiving CC data as well.
Thoughts?
Hi Debra,
Yes, there is a playback template (on Playback Template – Dorico). I did a search for Cantai and found it pretty directly. Though it sounds from your post like you already found it.
There were instructions on importing it and getting it set up correctly. Some users have had trouble importing it. Also, when using it in tandem with other instruments it had to be in a combined template but first in the list! There have been a number of posts about the problems users have been having in this area.
I don’t know about any expression maps yet. Cantai 2.1 reportedly did not do well with dynamics and there were already reports of 2.2 having its own problems! I think a lot of this is still being worked out!
Sorry if that’s not much help!
— Jim
To me it seems the playback template is useless right now except for selecting a specific space/stage. Since there’s no expression map it’s impossible to create an endpoint configuration and add that to a custom playback template.
Some quick tests show that the plug-in is responding to CC1 (modulation) but not cc11 (expression), I haven’t checked any others. So at the very least I would assume the playback template would include an expression map that utilize both velocity and modulation data.
I need to check the Steinberg website again, I may have missed something.
If you have the playback template installed, Dorico will automatically create Cantai instances when you add singers to your project, instead of having to do it by hand.
Yes, I’m aware of that (I have a number of custom playback templates that I use). Not having an expression map for the plug-in potentially loses access to features that might be already implemented.
So, Cantai responds to dynamic changes but is limited from pp-ff according to Richard de Costa, the developer. Personally, I think it should respond to the entire range of dynamics that Dorico uses but that’s just me. I also find that the choral voices are typically louder than the dynamic used in Dorico for other VSTs that I engage.
I’ve noticed some inconsistencies with Dynamics as well.
I went ahead and created my own expression map for cantai. I simply copied the default expression map and added CC1 as an additional modulator.
I do hope they release a more extensive expression map that utilizes all of the features within the plug-in. Also documentation would be helpful too! 
Cantai works differently for other plugins. Because it is AI, it has to prerender all the phrases, so it gets a copy of the “score” in advance of hitting play so that it can render before the user starts playback. Presumably this is the printed score and doesn’t include any CC data - I believe it actually looks at the dynamic symbols and hairpins as objects rather than following continuous controllers. Then the MIDI data is used to play back the prerendered audio in sync with the rest of the instruments that aren’t in Cantai (so MIDI mostly just used for timing information).
Originally the plan as I understood it was that Cantai was to get Dorico to export MusicXML and Cantai would process that MusicXML and use that solely for constructing the rendering and then play that back. But at some point that apparently changed to a different form of the score being passed to Cantai and not MusicXML. They still talk about Cantai getting “the score” as the rendering source rather than MIDI data, so I also assume it isn’t MIDI. This would explain why no expression map was mentioned or was used in the official template, because if MIDI isn’t doing anything except providing a timing reference, the expression map wouldn’t have very much to do and it almost wouldn’t matter what expression map was selected.
They did add the ability to do some shaping with continuous controllers, but to my knowledge it is just adjusting volume rather than actually giving you louder or quieter dynamic with the associated timbre changes. Any changes to dynamic should trigger a rerender of the audio.
Unfortunately a lot of the technical details of how this new method works seem to be getting glossed over at this stage, and I hope at some point there might be more clarity around that.
Yes, I understand the engine needs to read the entire passage to render the voices correctly. As long as the plug-in is reading the score, MIDI notes and all of their attributes, in addition to lyrics and dynamics it should have the ability to have all of the continuous controller data available as well.
Is there documentation for this plugin? I can’t find it on the website.
Thanks,
Debra
I don’t think there is any documentation yet. Although there isn’t much for the user to do at the moment aside from load the plugin and choose the voices that sing each line.
Hi! Have you had good results with your customized expression map?
To be clear, there is no evidence I can find that Cantai even reads the Expression Map so it’s completely irrelevant what’s in it. The dynamics are controlled entirely by the level in the dynamics lane plus Cantai’s own phrasing intelligence (which just occasionally goes completely haywire). I’ve experimented before with separate CC1, 11, 7 or whatever lanes and controllers in the EM and they are all ignored. The dynamics settings in Playback Options are followed but if you try to override them in the Expression map so that the voices respond differently, it appears to have no effect at all.
There really isn’t a need for an expression map at this point (from what I understand) so it was a fruitless effort to create one. There are no key switches, CC values, etc. to be processed by the plug-in.
I had hoped that there would be at least some fundamental controls of the plug-in. For example utilizing CC 11 for expression of a particular lyric sequence. Or program changes that one could change the voice of the singer in the middle of a score. Those sorts of things seem logical. But at this point the playback template simply sets the stage/space but does nothing else.
There is no expression map contained in the playback template which in my mind renders it essentially useless.
It’s also a bit irritating that the plug-in does not automatically start re-rendering any changes in the score. I’m not entirely sure what actually initiates the rendering, I’ve tried a number of things, such as changing the note pitch/length, adding a dynamic, or accents and staccato. It appears the plugin is still in it’s early days of development.
I was reasonably expectant that improvements would continue at a reasonably fast rate but it seems that all has slowed on the Dorico side in order to get the Sibelius version up and running. Perhaps Sibelius’ only recent ability to use VST3 has presented challenges to supporting both prior and current versions. I know a fair number of folks who did not upgrade to v26 of Sibelius (including myself). I don’t use Sibelius but did purchase the cross grade should a particular publishing house require Sibelius files for submission. So far, that has not been an issue.
Anyway, I’m getting a little disappointed that some critical issues (for usability) around pronunciation and sustained notes have not been already addressed. The issue that dynamics do not directly correlate to the range of dynamics in Dorico is another key issue that minimizes usefulness. I’ve played around with Compressors and Maximizers with minimal success for the final product. If the momentum doesn’t pick up soon, I’m afraid Richard and company may find themselves on the wrong end of competition that is working hard to catch up. There had been an approximate every 7 days update occurring from beta through launch.
As much as I strongly support small business approaches to development, I cannot help but wonder if a partnership with a larger development company might not speed things along a little …
Cantai at the moment has no more than a basic volume control built in which is designed to respond to the dynamics lane in the key editor but nothing else. Until Cantai itself provides the ability to shape the tone, I don’t myself see any need to put any extra controller support into Dorico. And I’ve never seen the point in a playback template for Cantai – it’s easier just to add the voices to the existing score as fas as I can see.
that’s not the case in my experience. Although there is a slight delay of 3-4 seconds, even the slightest change to the score I find is picked up and acted on.
I think which issues are critical is going to depend to some extent on the project. I’ve now converted around 3 hours of vocal music and have found nothing show-stopping and fewer serious issues than I’d expected. It’s true that the Cantai inbuilt dynamic flow means that lower notes tend to be quieter than high ones which is to some extent natural (and is emphasised a bit by pitch contour emphasis when enabled) but in the few places where the dynamics are clearly wrong, you simply add the correction manually in the dynamics lane. As for maximisers, I do on the whole tend to prefer the punch which the inbuilt Steinberg Maximizer provides.
There are some pronunciation issues but apart from really long held melismas and the occasional inexplicable exception, generally nothing that can’t be corrected by re-spelling words in the score. I’m referring to English here – other languages apart from Latin (which is not bad, I’ve found) are not supported and there’s no point at present using them for me. I found myself actually needing to correct the English more with SynthV which is a more mature product in general. If there are specific project issues other than the above you’re finding, I don’t mind having a look at a score if you like to see if I can do anything about them.
I’m willing to allow Richard and co. a couple of weeks to get the Sibelius version out (although I couldn’t care less about it myself, having removed Sibelius from my PC 18 months ago). And development has not completely stopped-- there is a new voice for Cantai coming out this week.