We have just rolled a product where we translate documents from DOCX to text.
The documents are in Arabic or Chinese.
Now, in production, the conversion often results in an empty text.
It seems to happen randomly - the same document is sometimes converted successfully, and sometimes it returns an empty text.
No error is returned - status code always indicates success, so there is nothing that could help with debugging.
Can you please provide more details about the method you are using for the conversion and any specific settings or parameters you are applying during the process?