I find that extremely creepy - for me maybe it’s an uncanny valley.
I get the same impression sometimes with Nina Simone - there’s something not right, not normal to my ears (this is a very personal opinion of course).
The trouble for me is it shuttles between uncanny valley and unintentionally hilarious with nausea-inducing frequency.
As I work strictly in pop music and theater, I wish there was some way to incorporate Synth V into it. That’d probably kick some azz.
You know that SynthV works great with Dorico? It will sync along with the score, but you’ll have to enter the lyrics yourself.
The big dream with software such as Cantai is the convenience of pressing play and hearing your score sung back to you. 
Just an update … Dorico is still NOT exporting XML correctly and my solution is still to ….
-
Compose in Dorico
-
Export to XML from Dorico
-
Open XML file in MuseScore (after clicking OUT of all the ADs … whoops, I mean, MuseHub “helper”!)
-
Export XML file from MuseScore
-
Render to Cantai online or AceStudio.
Dorico is not exporting XML correctly. Words (in Latin in my case) are not being recognized/sung correctly in Cantai or AceStudio. I can fix it (line by excruciating line) in the AceStudio MIDI editor … Cantai in its current form can NOT be edited/fixed. Exporting Dorico’s XLM export to MuseScore is the only fix.
Fix please! 
As from MSS v4.6, you can opt-out: Preferences>Advanced>Show occasional Musehub promotions. (agree, this should be an opt-in instead of a hidden opt-out)
Creep? Huh?
You’ll need to provide some more information about the nature of the problem with Dorico’s MusicXML export for us to be able to provide any remedy here. What, specifically, is the problem that you are experiencing with MusicXML files exported from Dorico?
I find there can be problems with any text which is not correctly aligned to the notes. For instance in a situation where text is copied from one part to another where there may be slight differences in the notation or other reason. I also find frequently that if the text is set as chorus then it is not recognised. In these situations the text is incorrectly allocated or often completely ignored. This is using SynthesizerV but in a casual test using the Cantai Dorico online renderer, Latin didn’t work at all – most of the text was simply missing.
However, if the above is observed, I don’t find a general issue with MusicXML export in Dorico in SynthV – and I have recently programmed several hours of music using it. I wasn’t aware that Cantai and AceStudio already officially support Dorico – Cantai is due in a couple of months with luck. At any rate SynthV does and perhaps that’s the reason why it generally works…
I think what dko22 is saying below may be the issue; meaning, if text is not aligned with the note “correctly” that the XML does not track well. My question is; how is it that the USER is not aligning the text incorrectly? Shouldn’t it be the case that whatever word/syllable you type to a particular note is tied to that note via a MIDI event? I mean, how could the USER do it “incorrectly”?
So, my use case is with BOTH Cantai & AceStudio which are new AI singing VSTs and it is understood that both are still in development and are improving every day and there’ll always be issues with using different VSTs with Dorico or any other App/Program (like Logic X, 11 & 12 and any issues that do arise with varying versions and updates).
Also, another issue is the XML issue. Now I do NOT know how XML works, but I do that with older versions of XML and with how the different notation apps/programs (used to) work, (I’m talking about MY experiences with Finale, Sibelius, Encore & Logic’s Score Editor) that XML worked differently and had different crashes/bugs that went with how each notation worked/exported with. Passport Encore, for instance NEVER exported XML correctly, so I would first do my draft in Encore, then export to XML (sometimes MIDI) and open in Finale, and finish/fine tune my ENGRAVING. Encore exported MIDI events nicely though (to Logic 8, 9 & X in my case), but XML never worked with Encore, while Finale exported XML the best out all them.
So, back to Dorico XML export. In my experience, when I would export XML to Cantai to import into Cantai’s online Renderer (which had to be done when Cantai first came out), the Dorico XML was hit or miss (for the ENGLISH), but the MIDI exports FROM Dorico have ALWAYS been ROCK SOLID, so when I exported a song in LATIN (MIDIwas/is required with Cantai’s online renderer to use lyrics in the LATIN Language), the Dorico exports worked. The only time Cantai’s MIDI/Latin export would have issues in the online renderer was a CANTAI issue, and not a Dorico issue. Sometimes the Cantai servers would time out or get stuck (which happens with everyone at some point), but logging in/out and restarting the browser would solve the issue.
Now that Cantai has the MuseScore VST version going, I now (as I did back my Encore>Finale>Logic 8/9/X days) compose in Dorico (ROCK SOLID, apart from when I jump the gun on a new release 
), then (for DAW/Recording) export to MIDI to Logic 12, and Dorico’s tempo changes, MIDI events export GREAT. Then I can record Audio Tracks to what I compose in Dorico with all ritardandi and key/time sig info, and it’s GREAT! (See the Screenshot). Dorico EXCELS at this better than any other app.
However, for the Cantai and AceStudio exports, I still have the export to MuseScore and THEN, re-input ALL of the lyrics as they get misaligned. Again, how can this be the USER exporting incorrectly if the manner that the user inputs lyrics is word/syllable to note?
In any case, foe XML exports Dorico offers “.musicxml” or “.mxl”, while MuseScore offers “Compressed .musicxml” or “Compressed .mxl” or “Uncompressed (outdated) .xml” files … With the different musicxml/mxl/xml options that MuseScore offers, at times a different version of the above will work BETTER with Cantai’s online renderer or exporting to AceStudio. Now AceStudio, you still have to fix/retype the lyrics EVERY TIME (ime) no matter what, so exporting MIDI from Dorico to AceStudio, or exporting to XML from Dorico THEN, to AceStudio while at times yield a better result. Again all these AI technologies (Cantai, AceStudio, etc.) are still in the works as we are all Crash Test Dummies to some extent, so you have to be patient and determined to (in my case) use the Cantai/AceStudio voices for only DEMOS (ie., working for a publisher/record company and getting them to use/buy your songs/arrangements and then use REAL contracted singers to be credited for a recording).
Back to Cantai … With MuseScore being the first to use Cantai as a VST, now NEW issues always arise (server issues, MuseHub/authorization issues) which is normal/expected. However, I still have/prefer to do all my composing in Dorico and when the song is ready for export for DAW/Recording, I still export to MuseScore, RE DO THE LYRIC INPUT as most of the time, for whatever reason, Dorico’s XML-lyric export is flaky. Then, I’ll wait for the Cantai rendering, and if all renders smoothly, I’ll export to Audio to Logic or Cubase (which I am still learning and WISHING there was a SMOOTHER integration with Dorico <<<< why the wait guys?!?!?)
I am hoping that with Cantai being released for Dorico next month (April 30, 2026 … it is March 13, 2026 today), I won’t have to do the painful, yet currently necessary XML export to MuseScore.
MuseScore’s LATIN sing-back/playback seems to have gotten worse than it was during the ONLINE RENDER ONLY option IME, and the online renderer has gotten buggy/weird; sometimes they exports are not of the AI voices singing back to you, but rather than it sounds as though the AI voices are having an asthma attack. Again, with Browser re-starts/re-log-ins, it’ll work.
Again, it must be how each app handles XML export, just like in my previous years of (as I said above) composing in the Encore, exporting Encore’s HORRIBLE XML export to Finale (which was the only program that accepted Encore’s XML) and I would engrave in Finale, and the export MIDI from Encore to Logic 8, 9 & X. Back then (and this is as recent as 2019 when I was still doing this), Finale handled XML with no issues as I recall, and Encore handled MIDI wonderfully. I am assuming its that old code that those older apps, I mean PROGRAMS had.
@dko22 , I would like to start using SynthesizerV, but the whole buy each voice is confusing and weired, and the voices seem more pop orieneted (a la AceStudio) which is great some songs, while Cantai works best for classical stuff.
AI voices are still a work in progress and the XML issue from app to app, you just have to keep trying. We’ll see. 
I accept in general terms most of your points. The first one about text alignment – if you type the text directly into Dorico (using SHIFT+L) , I hardly ever encounter problems. The thing is that in choral works, there is always a temptation to simply copy/paste text from one part to another and there issues can arise when the treatment is polyphonic rather than homophonic – in other words when the text is not simply vertically aligned. because Dorico’s spacing can understandably vary a bit from bar to bar. The other issue can be if the score itself is an XML import from elsewhere as some of my own were from Sibelius. I don’t necessarily notice the discrepancy at a casual glance but MusicXML tends to be more fussy in my experience. MuseScore’s xml export may be more widely compatible at the moment but I don’t see that means Dorico’s is intrinsically flawed – probably Ace and Cantai have been more widely tested in MuseScore. When released for Dorico, Cantai will, I’m sure have been tested with Dorico’s MusicXML export.
As regards SynthesizerV, I use the new choirs only and choir 2 is in particularly very suited to classical works in my view, given appropriate tweaking, as I tried to demonstrate in the various examples I posted on the made-in-dorico subforum. If you find even they sounds too pop-like, then you might have a long wait as Cantai in its current state in undoubtedly worse, even if it’s aim is – laudably – to cater for the notation/classical crowd which will include the bulk of Dorico users. They have promised a choir of their own and, like many, I’m interested in what it will bring to the table.
Yes INDEED dko22! I had many more issues/problems when I would copy/paste the lyrics initially, but I figured that out right away. In Dorico, each of my projects has multiple flows for each song;
-
Condensed Score, for publication with all instrumental parts and each part has their own version of the Condensed score with (the necessary) instrumental queues. The Condensed score has the SATB/Choir parts in a Grand Staff (where applicable; sometimes a Descant w/ an additional staff) & repeat bars, numbered endings, etc.
-
THROUGH Score, for conducting with all instrumental parts and each part has their own version of the Through score with (the necessary) instrumental queues. The Through score DOES NOT have repeat bars, nor numbered endings, etc, but does also have the SATB/Choir parts in a Grand Staff (where applicable; sometimes a Descant w/ an additional staff).
-
DAW/RECORD Score, for DAW export also with the same instrumental parts and each part has their own version of the Through score with (the necessary) instrumental queues. The DAW/RECORDING Score (like the Through Score) DOES NOT have repeat bars, nor numbered endings, etc, but does also have the SATB/Choir parts in a Grand Staff (where applicable; sometimes a Descant w/ an additional staff), IN ADDITION to the condensed SATB staves, I’ll add a staff (and it’s corresponding MIDI Channel/instance of Olympus) for each of the SATB voice (Sop, Alto, Tenor, Bass, Descant, etc.) and create a particular layout for that called “Choir Only” which has only the Sop, Alto, Tenor, Bass, Descant, etc. voices (each with their own MIDI Channel).
With the “Choir Only” layout, my workflow is as follows:
copy the SATB Refrain TOP STAFF (Sop & Alto) only which highlights only the Sop & Alto and lyrics …. then
Ctrl+Click>Filter>SelectOnly>Voice>All Up Stem Voices (which is the Soprano only)
Ctrl+Click>Filter>De-SelectOnly>All Lyrics so ONLY the soprano NOTE is selected, and I’ll copy the NOTES ONLY to the new Soprano only staff, then I’ll re-input the REFRAIN Lyrics for the Soprano part, often times typing in phonemes for words like Alleluia so it will SOUND like “Ah-leh-loo-yah” rather than “Aw-LAY-LOO-YUH”!….
I’ll do the ABOVE for each of the other parts (Alto, Tenor & Bass, and descant where needed).
I’ll the same process for the verses (where applicable) and, since this is a through score and Cantai/AceStudio can’t sing a line with multiple lyrics, I’ll RE-TYPE IN ALL VERSE LYRICS FOR EACH VERSE.
So, the CONDENSED SCORE will have an SATB Grand Staff with the Sop & Alto on the upper staff, and the Tenor & Bass on the lower and the VERSES will lie under each note with each additional verse lyric under the previous <<< this is for ease of use/less page turns. The THROUGH SCORE requires pages turns for each Refrain, Verse, etc. The DAW/RECORDING Score has all of the Through Score stuff, but is meant for exporting.
So, I HAVE DONE MY WORK and GO THROUGH THE SUFFERING TO MAKE THIS WORK! LOL Even with all the above, the XML/MXL/MUSICXML (whatever is used in Dorico) from Dorico to MuseScore>VST Cantai, or I’ll do an XML export each of the Layouts of Sop,Alto,Tenor,Bass from Dorico (so each has their own XML part) and I’ll use the online renderer of Cantai or I’ll drag the XML exports of each Sop,Alto,Tenor,Bass layout from Dorico into the TimeLine of AceStudio …. each time, it’s a crap shoot on which works better. Sometimes the Dorico XML export (of each part, or the “Choir Only” Layout) will work, sometimes it won’t. Sometimes the Dorico XML export to MuseScore XML export to Cantai/AceStudio/Cantamus will work better. Sometimes, not. I’ll get errors, mis-pronunciations, etc.
Again, like my Encore XML to importing into Finale working better back in the day, sometimes some apps/programs handle XML differently. Some handle MIDI better.
So, I’ve done all the steps to get the results, it’s the bugs and issues that ALL apps have. No hate to Dorico, which many get VERY sensitive and nationalist about. :-/ We’re here to report issues and solve them by sharing knowledge. I hope the above clears things up.
Regarding SynthesizerV …. might you be able to share which particular voices are the “…. new choirs only and choir 2” as I would like to try those out. I am not keen on SynthesizerV’s “pay for each voice” way of doing business, which has turned me off from using them, but if those particular voices work good for Choir/Classical stuff, I’d love to know which ones if you wouldn’t mind sharing. Thanks a bunch!
Oh by the way … Church music isn’t only in LATIN … Kyrie eleison is GREEK … Alleluia (Hallelujah), Hosanna, Amen, Emmanuel are ARAMAIC/HEBREW … None of the AI platforms pronounce them great at this point. Kinda like a real choir LOL, but with a real choir MOST of the time, you can correct most of them.
However, hearing “Aw-LAY-LOO-YUH” (very CAAAAALUFORNYUH!!! <<<I direct choirs IN California, believe me, I KNOW!!!LOL) instead of “Ah-leh-loo-yah”, for the word ‘Alleluia’ or “Uh-mah-nule” instead of “Eee-mahn-yu-well” (for classical English Speakers) or even “Eh-mahn-well” (for Spanish speakers) for the word ‘Emmanuel” … these can be maddening! AceStudio’s interface allows for adjusting phonemes and syllables and AS and Cantai do allow for entering phonetic syllables, with VARYING results, but the day I do NOT have to type KIT-TY YEH EH LEH EE SONE is the day I can die happy!
Non of the AI Voice platforms (not Cantai, nor AceStudio, nor Cantamus, nor SynthesizerV) have Greek and certainly not ARAMAIC/HEBREW so getting your AI voices to sound “Chanty/Churchy/Cathedraly” rather than like Selena Gomez or Tayla Swiffs. (No hate to pop music, I do that too, and that is where AI shines and is marketed to).
This is where adding a LAYER of Voxos and OPUS Builder (Symphonic Choirs) really helps. That rounded, “cotton ball in your mouth” kinda singing actually works in OPUS & Voxos when using phonemes. Again, why all the work for such slow results?!?!? 
I know what you mean but I’m not among them. I use Dorico simply because it’s the best tool for the job, which doesn’t mean I’m above pointing out things which I think could be done better. I have to say I couldn’t imagine going through such a tortuous process for vocals – I simply write the music and export the MusicXML.
Every one of my works uses a different configuration. I use two choirs and in each you can chose anything from a solo to 16 voices per part with a choice of four soloists for each voice. If you find a work from my output where the voices seems to work for you – at least to some extent – then let me know and I could share the SynthV file so you could see the configuration.
well yes but 99% is Latin. I find that SynthV Latin using the European Spanish model is fine, even if there are obvious corrections which have to be made. Not surprisingly the general Latin sound is not dissimilar to the Catalan Cantamus. American English is pretty hopeless as a basis for just about any other language – for German, Swedish and Russian I have also used Spanish for want of anything better at the moment.
I had actually wondered about this! There are certainly situations where the actual sound from the Hollywood Choirs in particular is preferable but I suspect trying to combine them successfully would be beyond my mixing skills at any rate.
Guess it depends on what you mean by ‘church.’ There are many, many churches who’ve probably never HEARD of Latin, let alone used it.
Well of course I – like @daniel_delago, presumably – was referring to sacred choral works which are nominally in Latin. Few churches have Latin as the actual language of a service these days.
The point I was trying to make is that ‘sacred’ combined with ‘choral’ is not necessarily connected to one particular religion, in this case Roman Catholicism. There are many ‘sacred choral works’ connected to other religions and in other languages: consult Brahms, for just one example. German and Lutheran. Or Vaughan Williams or Britten: English and Anglican.
‘Church’ is a big word. ‘Sacred’ is even bigger.
obviously – I don’t see any disagreement here. The discussion was simply about the fact that the Latin Mass includes a small amount of text in other languages. It’s not about other sacred choral works - even I have written one in English! And happen to be currently singing Ein Deutsches Requiem to which you just referred. To say nothing of the many other religions which have their own vocal traditions.
And Latin isn’t even Latin.
There is the Italian pronunciation where c before i and e is pronounced like in modern Italian used in works like Verdi’s Requiem, and there is the German pronunciation where it’s pronounced like ts or z used in Bach’s and Mozart’s Latin works. Other differences are written sci in suscipe and gn in Agnus.