General feedback

I live this one "song" called "Text-to-speech Models Suck" showcasing the possibilitys of udios good text to speech model, I've tried to remake it for myself in German, translated the text, but wasted 400 coins, and got a bunch of nonsense, gibberish and a terrible voice, this is also for "music" what udios is intended for, I know. German "singers" almost always sounds robotic and like a Bored student.

When using custom lyrics, occasionally when you want to edit it and hit backspace/delete the text is put out of alignment and squished. From what I can tell, it seems to happen after you back out of the create page and then go back into it in the same tab/instance. This is extremely irritating to deal with, hopefully this issue can be expedited up.

When creating a new song or making a song by pasting, language changes often occur. It is rare for a country's language to appear consistently until the end of the track, making it almost difficult to use. For reference, lyrics that were in Korean often change to Japanese in the middle. Is this inconvenience only happening in a specific country like Korea??? Please improve it so that consistent languages ​​can be created!!