What is discord channel… your link brings me to a login screen

Yes, it's the community discussion board. Cantai don't provide support there, any more than Cantamus do on their Discord channel. It'll only cost $10 to try Cantai for a month to see does it work with your specific music and desired voice types. You'll probably need to sign up (free) to their Discord channel to post problems and get the workarounds from those who have been there before.

I have it (was an early supporter for 2 years before a product dropped) but as of yet don't use it much. There are still rough edges. The good news is that development is happening at a breack-neck pace, and it is getting better. Early iterations had vibrato that would have made a billy goat blush. Recent choir voices have toned that way back (thank God). But there are still issues being worked out.

Dorico

@francis, I suggest you put that point on their Discord channel and see if you get any ideas to try. I'm not following it closely but people are constantly coming across and overcoming limitations and where that fails submitting bug reports.

This is for playback and notation software purposes, which is not a religious use

A friend of mine pumped one of my recently finished scores through Cantai. Hearing "actual voices" sing it (cue AI debate) was crucial for the next step of editing. It's a powerful tool that I would highly recommend for the composing and editing process. I would not put the sample it produced out there for the world to hear because it was *bad,* like, a sloppy and sometimes drunk-sounding. However, the consonants and vowel noises made a world of difference listening to my score. So for personal use, great. For good sounding recordings, not a chance.

This is a great tool... I have just scanned in all the PAPER (with ink or pencil) comps I have composed in my lifetime... (have everything back to when I was seven years old) (see score attached).

It took three weeks, eight hours a day to get it done. 98% of them have never been performed. This tool will let the world hear a close simulation of these works. It really is an audio documentation of what exists on paper. This does not include all the works I composed in Sibelius which is probably also a thousand or so. The ones that are in Sibelius can now be turned into audio simulations so people can hear them too.

We (composers) would never promote using AI to replace real voices. God wants 'worshippers in spirit and in truth'. That comes from a human 'heart, soul, strength and mind'. That happens through lungs and voices supported by minds and hearts filled with love through dulia, hyperdulia and latria.

@ServiamScores

Nice!

I hail thee...

http://www.myopus.com/preview/sitio2026.mp4

One of the settings of Seven Words in my arsenal premiered in 2009 with string orchestra and choir.

I've been composing for just over a year, always in MuseScore. As an adult learner of violin and bass viol, with a few years lessons but otherwise self-taught in both playing and music theory, I find composing in MuseScore an excellent way to learn about making music. I've written some things that would have been impossible for me to play a year ago, but because every note is instantly rendered as I type it, my ear has led me to write things that have forced me to become a better player, such as shifting higher on the violin, or rhythmic challenges that hearing the machine-rendered phrase teaches me to play what I couldn't quite sight-read.

Now about Cantai. After learning in this discussion that Cantai exists, I did some online reading and decided to get it. I've never paid for any MuseHub sound libraries but Cantai is included in the MuseSounds Pro bundle so I paid $149 for a year's subscription, and spent most of yesterday trying to get Cantai to work.

WARNING to anyone considering Cantai: to properly run, Cantai requires certain processor chips in your computer. For example, on my 2020-era iMac (that runs perfectly, including every MuseScore use until yesterday), the intel core i5 processor will not handle the rendering of the vocal notes on the score into audio sounds in an audio file. So Cantai reverts to an online processing by their server, which takes many hours and often shuts down (giving "check connection" notifications despite my internet running fine and being logged-in to both MuseHub and MuseScore). After all day yesterday the online processing continues today, a rather brief score still incompletely rendered, so upon reaching the end of the rendered part, the vocal notes sound like a piano rather than sung words. So, for a viable work flow, do not pay for Cantai unless you have first verified that whatever processor is in your computer is suitable for "local" (in YOUR computer) rendering of the sound by Cantai. I have never experienced my computer processor being unable to run a program that is compatible with my MacOS, so I did not even know to check whether the processor will run it.

However, those passages that did successfully render via the online processing are, except for some small glitches, very beautiful! Real words sung instead of oohs and aaahs, so consonants and vowels give a realistic sound in playback and therefore better feedback for me as a composer trying to decide if it is right or needs revision. Just hearing those glimpses makes me want this Cantai software to really work in my MuseScore sessions --so much that I am already thinking frivolously of upgrading to a Mac that has an M-series processor instead of an old intel processor. If I were rich I would have ordered the new computer last night already, that's how much of an upgrade to vocal rendering Cantai is compared to the free MuseHub sounds I'd been using until yesterday.

Funny enough, the other libraries like Vienna Symphonic or Spitfire instruments I did not like more than the free instruments I was already using, although in different scores I will have different instrument voices and might revise that opinion as I explore those libraries.

Well, I have to say, my initial skepticism has been put in its place. The updates of the last few days have really resulted in a marked difference from the last time I tried it. I arranged Remondi's O Sacrum Convivium into English and made English and Latin demos of the piece. WOW. wow. These aren't perfect by any means, but entirely passable, and perfectly adequate for a practice track. I'm gobsmacked.

Latin:
https://youtu.be/8QvFKmQEeDI

English:
https://youtu.be/bxkkupsACrk

Links to free scores of both the Latin and new English adaptation:
https://catholicsacredmusic.com/blog/lets-celebrate-5k-free-remondi

The last time I tried it was a few months ago, and it was still terrible at that point, IMHO. The early releases had vibrato wide enough to drive a barge through. (the developer is an opera man) thankfully, he reversed course and recorded some normal singers (lol). There were many more intonation / pronunciation issues earlier on as well. These have also drastically improved. Now with their otteto voices (renders more than one person per part) It's genuinely passable.

I'm on an m2 mac pro with 32gb ram running Cantai in Dorico.

That is terrifyingly good. We used to think recordings ruined the ability to appreciate live performance. AI voices will sideline new recordings because recordings will be too expensive to produce.

Francis, “prefer” is also the word I would use. But as the kids grow up into a world we’re AI voices are ubiquitous, I think human voice recordings of new music or new recordings of old music may become just an expensive preference. Sort of like collecting and preferring vinyl in 2026

as the kids grow up into a world we’re AI voices are ubiquitous, I think human voice recordings of new music or new recordings of old music may become just an expensive preference. Sort of like collecting and preferring vinyl in 2026

I absolutely hate contemplating this...but unfortunately I could easily see it happening.