When I compare between visual novels that are originally developed & written in Japanese first before localization into English, often or not: the information density is different between the two. As in, the original text only comprises of like 2-3 sentences while the translation feels like reading a 4-5 line paragraph in English. Is it because translators who managed to secure the rights can’t cram all the information in a limited UI?
For example, Aksys Games is the one who does the translation while the IP holders from the source material is Otomate. They only focus on the text rather than re-doing voice acting for the English release. Lip sync is another issue since both JPN & ENG have different speech patterns, how would you even find the right casting pool in English that can deliver the same performance & emotional impact as the original JP voice actors?
As for lip syncing, I have always thought that translation services, particularly for animated works, should be able to adjust the mouths of the characters for the purposes of adjusting lip syncing. Different languages have different cadence, speed, etc.
You could just not care about lip sync, like old Godzilla and kung fu theater type films, but for animated works it seems like the editing would be pretty easy and minimally invasive. It could free up the lip syncing for more accurate translation. At the same time, the facial expression would still need to match the original intention of the character.
I recently watched a series that used some kind of generative AI for lip syncing in the English dub, as the show was originally Japanese, but it was a live action show. In theory, this was a pretty good use of AI, at least to me, but in practice it ended up looking really weird because the mouth movements didn’t match exactly the voices AND the lighting was slightly different. It ended up being really distracting in the end. The technology isn’t really there yet for it to be convincing for live action, but I don’t see an issue for this for animated works.