I’ve encountered an issue after using AI pre-translation in Crowdin. While most of the translation looks fine, I noticed that some translated strings appear 2-3 sentences later instead of aligning with their respective source strings. This results in mismatched translations in certain parts of the text.
Well, it might depend on various factors. Do you run pre-translation on the strings level or send the whole file to the AI Provider?
Usually, if anything wrong happens during pre-translation, there’s a description of what went wrong in the project tools (service logs). You might try specifying in the prompt to go string by string or to ensure nothing is missed.
To fix the already received result, you could try the following steps:
Identify the strings that are out of alignment.
Delete the incorrect translations for these strings.
Use the pre-translation feature again, ensuring it’s applied only to the untranslated strings.
This should help in aligning the translations correctly with their respective source strings.
If the problem persists or if you need further assistance, please provide specific examples of the affected strings.