For choir members it’s a VERY useful tool, especially for people who don’t read music or have little time to practice, they can even practice in the car. I export the mp3 for each part and for the whole choir together and distribute them to the members. Really happy with this tool.
I don't get it. What do most of the voices sound like very sickly little children?? There's only 1 useful baritone, and none of the tenors voices are useable. The so-called "opera" voices sound as if they're singing while on life-support. The mezzo choice are reasonable, and the "Rigoletto" baritone are useful. But beyond that...WOW. Where is the real Cantai quality choices?!?! Please FIX THIS! we're paying for this product which is still struggling.
Is it perfect? Does it save real singers? No, absolutely not. Is it soooooo much better for writing and arranging songs for voices and choirs than anything available so far? Oh yes!!!!! As someone who focuses on vocal arrangements, this is already a gamechanger!
Version 3 seems a lot better. I really REALLY hope it does not suddenly disappear again... Actually I tested Spanish and it is pretty good, could be part of a WIP but it's working for me. Muy bueno.
Its a cool idea and it definitely has potential, but right nowit needs a lot of work. A lot of the voice tones need work,most of them sound kind of choppy instead of smooth and natural like a real voice. The only one that sounded close to natural is Aurelia. The biggest issue I ran into is timing. When you use the AI synth, the vocals end up delayed, so they don’t line up properly with the instrumental parts. I did find a workaround though: I exported the vocal line without Cantai, then exported another version with Cantai, and lined them up in a DAW. That works, but it really feels like something that should be fixable inside MuseScore itself. It would help a lot if there was just a simple track delay setting so everything could be aligned properly. It would also be nice if you could some how control what notes you want vibrato on. also noticed some dynamic issues, but those are easy enough to fix once you move things into a DAW. Overall, it’s not quite there yet, but it shows promise if the timing and vocal smoothness get improved.
It's kind of nice to be able to listen to a score with the lyrics actually sung, but that's about as far as I can praise this. The voices sound warbly and poorly-tuned, especially using the choir (male) and choir (female) options but I'm also having the issue with solo voices too. I also find it hard to get the voices to actually sing the word sounds that I want them to - making it sing "live" with a short i sound proves difficult, and for some reason it just forgets to sing some short words like "I"'s between other words. The lack of information about what the voices will actually sound like isn't ideal either - "Marco" and "Carmen" mean nothing to me. The official Cantai website only has information about a couple of these voices, with some on MuseScore not even being mentioned there. I'd much rather just have them named "bold soprano" or "soft bass", rather than these meaningless people-names that just seem to be there to try to make you forget this is an AI voice. The website claims that they're using ethical AI and that the singers whose voices are being used are getting a cut, but the fact that several of the voices don't have credited singers, and many are missing entirely, makes me unsure on that front. All in all, I'm glad I only got a 1 month subscription. It's far from good and I'd honestly rather just listen to my scores with the basic MuseScore sound pack, and sing it myself.
The only thing decent about it is indeed the text pronunciation, the vocal samples sound very bad, the terrible vibrato in the make and female choir is very off putting, so you are left with the solo voices that sound like the only repertoire they are used to is Amercan Christmas carols and even the intonation is off, the dynamics are off. I'm glad I only bought a month subsrciption. Worst thing is, the syncing with online voices gets stuck most of the time during processing and it's not my glassfibre internet connection causing it. So no, not better than nothing, The cinesamples library has way superior samples but unfortunately no lyrics pronounciation.
Are there plans in the near future to add voices more suitable for country and southern gospel genres? The existing voices are a bit too classical/operatic for that.
Voice tones sound very bad, the intonation is not accurate at all.
This is certainty better than nothing but there's not enough here just yet. You can't change the ensemble size and this sounds like a vocaloid. That being said, it's early tech and is impressive to get on musescore of all things.
it works reasoanbly well, but it is missing more normal tenor and bass voices. A big issue is that every time even a small part is changed, the voice on the whole score is updated, and this is slow and only online. This makes it hard to use with bad internet.
I wanted to try this on a musical I am writing, but was sadly disappointed to see there was no classical tenor voice. I was also a bit disappointed there wasn't a more classically trained soprano. The choir voices are perfect, but they don't work in soloist settings. Overall, I really like the product, but it just needs more to be worth the price of admission. EDIT: Cantai just dropped a huge update with a whole plethora of new voices, and wow! I wrote a lovely little choral piece using them and was very satisfied with the quality. I have yet to try the musical theater sound libraries but so far so good!
If you want any type of classical sound these are AWFUL! Way too much scooping and wobbly vibrato. Solo Tenors - not one decent sound. Baritone only has one halfway good sound. The choir sounds are atrocious!
I highly appreciate how easily it generates very realistic vocals. I'm giving it a high rating to show my support for future development. I'll mention three points I'd like to see improved. First, I'd like the ability to control where breaths are placed. Currently, even if a breath mark is on the sheet music, it sings without pausing for breath regardless. It would be better if it either automatically pauses for breath or allows me to specify the breath position myself. Second, I'd like the ability to control singing techniques like vibrato depth. Especially when creating sources for early music, contemporary music, or educational purposes, excessive vibrato can be distracting. It would be good to specify singing techniques in addition to the singer. Third, regarding the unnatural timbre changes accompanying dynamic shifts. Currently, even when transitioning from fortissimo to pianissimo, the timbre remains unchanged—the sound simply gets quieter as if someone turned down a volume knob. This feels very unnatural and inhuman. I'd like the timbre and singing style to change naturally when dynamics shift. Overall, it feels like all options have been sacrificed for ease of use. Technically, I believe this is entirely feasible, so I look forward to future development.
UPDATE: The new 2.0.1 version is causing sound processing problems I didn't have before. Let me start with a summary by saying I'm glad I have this available, but there is a lot that could be improved. The simple fact that I can now hear my works with the words sung instead of choir "Ah"'s is wonderful. Having said that, the choir "Ahs" of MuseSounds are much more pleasing and choir sounding than the choir voices of Cantai. The Cantai "choir" reminds me of the background vocals from commercials or back from the crooner days, except not as good. Since my main medium is choral music, this is disappointing. It sounds like the choir is a collection of the solo voices, most of which are weak and very AI-ish (especially the male voices). Unless you are going for the untrained high school voice sound or a wanabe Karaoke sound, you don't hear "Choir" with these voices.. I would think you all would have the means to create the pleasing "As one voice" sound that the Musesound voices have. There are a few decent solo voices, which makes me wonder why the others are so weak. Most notable is the lack of a respectable tenor voice. Although less frequent than with ACE Studio, I have to re-spell some words in order for them to be pronounced correctly. Given the complexities of the English language, that is understandable.
Please increase the processing speed!!! I keep it on maual processing, otherwise it would drive me crazy. I like the realism of the voices, although alot of them are too stylized. There also need to be a true bass and alto, instead of just baritone and mezzo soprano.
takes forever to process then crashes my musescore. doesn't save after the crash. come on guys im tryna write some bangers here
I really like the choir voices overall. The soloists are hit or miss.
Probably the only positive outcome of machine learning technology! I have used this for my students who have little experience before a concert as an aural guide. Thanks for this! I wish it could be a single time fee, like the other major notation software, but I will take what I can get!
The update brings new solo voices compared to 1.0.0. However, the new voices don't sync up in time with the original voices. So in a SATB setting, new voices and old voices are not compatible. I hope this can be fixed quickly as it does affect the functionality. For record, the original voices that are in sync are: Lucia (Soprano), Karolyn (Mezzo), Giovanni Opera (Tenor), Rigoletto (Baritone) Edit: Hi Cantai team, there is a misunderstanding in the response given. I am not talking about menus and reassigning, I am referring to the actual time words are being sung by the various voices. The new and old voices are off by around 200 milliseconds, which causes a real mess when played back together in an SATB setting.
All of these voices are gorgeous! I find myself using them a lot for dramatic or thematic orchestral works. The sound quality is tip top, but there are a few bugs I must work around. One is sometimes the lyrics I type don't match the sound, so I have to artificially get it to resync by removing a word, then readding it, or sometimes I must just remove the word entirely. Also, for a song one of the main words was "Tears" but it registered it as the "tear something apart" one. It would also be very useful to control some of the sounds. For instance, in a song the singer at the end of a dramatic high note kind of did a glissando squeak up, which sounded perfect, but the next time it synced it got rid of it. Just a few things to keep in mind, it's overall a great product and I'm excited to see the future updates! --Edit-- That all said, the Men and Women choir sound very unrealistic and harsh. I think what you tried to do was combine and meld already recorded solo voices into a choir, and that never works. I would highly recommend you record a real choir. I know its expensive, but a real 50-person choir would be awesome and it would easily take a spot in my orchestral palette. Thanks!
As a choir director, this is awesome. I'm just a bit bothered by the fact that Latin text is sung with English pronunciation. And I'm curious to know when will we be able to have Italian and German pronunciation? As almost 80% of the Western choral and solo Voice repertoire is in Italian or in German. Or even in French
I've been waiting for this for awhile, so it's awesome that it's finally hear. I write hymns and score public domain hymns to make them singable, and this helps tremendously. It is so much better than the old "oohs" and "aahs". A few issues/features that I hope will be addressed: 1.) I wish I had the ability to have the program select which verse I want it to sing. As it is, I have to score the music. Put in the lyrics for verse 1. Export the mp3. Delete the lyrics and put in the lyrics for verse 2 and rinse and repeat. It's doable, but a pain. 2.) I hope the program does a better job of singing shortened words. It really struggles with heav'n, and forgiv'n for example in the score I'm currently working on. It seems like when it doesn't know what to do, it just gives sings basically "aahs". This is really annoying as one can probably guess how often words like this show up in hymns :) Regardless, thumbs up for the program. I know it's a work in progress and progress will be made. Thanks!
As a classical composer with great grandchildren and very little money, the fact that my low-end laptop can sing my music brings me near to tears. I could quibble about a few things, but it seems as though the developer has plans that will address those minor (to me) issues.