Works pretty well, except it doesn't always interpret vowel sounds correctly, especially for words that have two meanings and pronunciation. It doesn't seem to understand the context. For example it read the word "lives", which I intended as a verb, as a noun. Vowels can be edited to add the phonetic symbol over the letter but that is a bit cumbersome.
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.
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.
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.
It's amazing how tools are improving every single day. I've been looking for a tool like this that's compatible with Musescore and I don´t have to export xml, or cut parts, etc... It's kind of dissapointed the only laguages available are english and latín. My native language is spanish and also Irecently write something in italian and I wanted to give it a shot, I gave me an idea but of course with english accent. Also I try with some english songs and it was much better but still need to fix some bugs bc is not recognizing some words... Also, I couldn't export an mp3 file, to show a sample to someone :( I think is promising and definetely I'll keep an eye on updates, but at the moment I don't really think it is worth to pay a monthly subscription at the moment, but if we improve the languages most likely I'll be there :)
At the time of this software being released, I just so happened to have completed a choral symphony, and since I unfortunately don’t have singers at my immediate disposal (as with a lot of composers, I’m sure), it’s nice to get some immediate feedback regarding the vocal music composers write. While I’m sure that the voices will be improved upon in due time, my only main gripe is the monthly subscription; I had actually preordered a one-time paid lifetime subscription, so that I wouldn’t have to worry about putting aside money every month; this is quite detrimental, especially for the younger composers who have a specialty in choral music; other than this though, it has delivered quite beautifully for me and I can’t wait for what’s to come in the future!
I’ve always wished for a product like this, so in a way it really caught my attention. Unfortunately, it doesn’t quite deliver. The biggest drawback is the monthly subscription. I might understand a one-time payment, but paying every month for something of this kind simply isn’t worth it. Another problem is the soprano range, which is surprisingly poor. And I can’t overlook the English accent present in all voices. These issues make the product feel unfinished, almost as if it wasn’t tested with real musicians in mind. If these shortcomings were addressed, I genuinely believe it could become one of the best-selling products on MuseScore. But if the developers don’t fix them, I’m certain a better competitor will take its place soon enough.
It’s a very cool product that will be very useful for operatic or classical scores. Unfortunately, as a musical theatre writer, these voices aren’t helpful to me at all. This product will be a 10/10 if/when musical theatre style voices are added!
Impressed by the skills putting this library together. Works very well for opera scores and the opera Cantai voices but less so for music theatre, musicals and pop (the soft and ordinary voices of the tenors and barritone are not great). Unfortunately the soprano is pitchy and strained in the upper range - Ab is not high enough for most opera scores and sopranos can hit this note without strain in the realworld. For sustained notes over multiple bars, the pitch dropped in the final bar before scooping back up to the correct note (for any of the tenors) and there are some mid-range notes that are also pitchy. Not enough choice for sopranos, mezzo and no alto or bass (hope these are coming). The online sounds have to be reloaded every time I open MuseScore - not sure about the cache notes in the manual - doesn’t seem to be the case for me as when I first open score that has Cantai already used, the default piano sound is active - using MacOS 26.1
I wholeheartedly agree that this product itself delivers exceptional results, resolving numerous issues with vocal-heavy sheet music that previously required human intervention. I am personally acquainted with several opera singers who work with this product. However, its limitations are significant. Firstly, it operates on a monthly subscription model, which creates considerable uncertainty for users. Most critically, it imposes a weekly quota of just 180 minutes. Regrettably, there is no transparency regarding how these 180 minutes are calculated, meaning users can easily exhaust their allowance when initially testing and adjusting the product. After weighing the audio quality against the cost-effectiveness, I ultimately believe the developers need to offer a more reasonable usage plan that benefits consumers. I fully acknowledge that MuseScore has consistently been a latecomer achieving technological progress, but careful consideration is required for specific market strategies. For this particular product, I personally feel that removing the quota restrictions would be the optimal solution.