希望能让跨语种功能的发音更加本土化和标准，这样能充分的提升声库的可利用性
Xingchen english problem:
“glare” /g l eh r/ becomes “gwear”
Also a problem with “gleam” → “gweam”
Glow sounds fine though, english native VBs seem to have no problem with these words.
As a workaround, you might try using the single phoneme /gl/ as a substitute for /g l/. Sometimes one works better than the other.
Ryo AI has issues with terminal /r/ sounds.
For example, “air” is /eh r/, but there’s just about no /r/ sound, even adjusting the parameters. It’s sort of a soft British /r/.
「いいね！」 1
With Kevin, there’s a bit of devoicing going on with the /g/, which makes it hard to distinguish from a /g/.
星尘AI的pian变成了kian(中文),有互换现象
岸晓-n结尾的短字有bug。在句尾还好，但在句中会有可能听起来像另外的字。若后面连着字，似乎会（浊化？）改变辅音。除了power声线，其他声线（包括normal）都存在这个问题。
例子：
zun yan，C4，1/2音符，听起来像zui yan，无法通过发音长度/力度/alt发音/音区偏移/改变发音符号改善
zun chuan，听起来像zun ran
zun nian，听起来像zui lian
I’m not sure this is the solution people make it out to be. Even within English, there are still regional dialects.
Listen to voices that use cross-lingual synthesis, and you’ll hear that there’s more to singing in a language than matching phonemes sounds.
「いいね！」 5
nah, “to singing” is right in this context. You would need a noun in that context, so it would either be “There’s more to singing”, or “There’s more to song”
Examples:
There’s more to song than just hitting the right notes.
There’s more to singing than just hitting the right notes.
*There’s more to sing than just hitting the right notes.
You could say “There’s more to sing.”, but not with “than” and it means an entirely different thing.
EDIT
So, X-SAMPA unification might be useful from a phonetics standpoint, it will make the phoneme specification further away from an individual language, which should be taken as one wills.
「いいね！」 1
X-SAMPA would be beneficial for languages that aren’t officially supported, and for multi-lingual songs.
But the trans-lingual feature is Dreamtonics process training a neural network target phonemes in language that weren’t recorded by the voice talent.
But since the appeal of X-SAMPA is that it moves away from custom languages, it’s sort of the opposite of trans-lingual.
「いいね！」 2
Changing English and Japanese to use X-SAMPA would bring minimal benefit since the actual available phonemes would not change, just the symbols that represent them.
It is important to note that the simpler representations currently used are much easier to learn, and the more intuitive mapping makes it much easier and less frustrating for new users to pick up the software. Using X-SAMPA for all languages would introduce an unnecessary barrier for new users.
…also are you really suggesting that Kanru chose the various phoentic alphabets based on how much he likes each language? 
「いいね！」 5
That is not what I said. Your condescension is wholly uncalled for.
My point is that if you want to change something, the benefits of the change must outweigh the disadvantages. You have yet to provide any concrete reasoning that using X-SAMPA would actually improve the product in any meaningful way.
Jumping to wild conclusions based on things I never said is not going to help your argument.
「いいね！」 6
ISSUE: when I use Kevin in Japanese cross-lingual, his /k/ and /t/ are pronounced super violently and cut off the previous vowel, almost Eleanor Forte-like. i always have to go and adjust the duration and power of the phoneme.
I presume you mean: Hard to distinguish from a /k/.
That’s gorrect! 
「いいね！」 2