Question about translation packages to translate websites
Hello,
support.mozilla.org/kb/website-translation says, when you do not have downloaded translation packs, parts of the packs will be downloaded when you first use the translation function.
My questions about this feature: - How does this work exactly? - Will the words from the website which i try to translate be collected and "downloaded"? - How are the translation language packs parted? - What information will be used and sent to servers/cloud to request these parts of the translation packs?
Thank you in advance for your help.
Best Regards
Bjoern
所有回复 (4)
Hi, this support article is not accurate, in Firefox we support only complete models.
However, Gecko engine can handle so called "incomplete models", but it only refers to models that hasn't been fully downloaded due to error. We don't generate partial models, nor we have it in a download storage.
This is confusing for me. How can Firefox then provide a local translation although i have not downloaded a translation language?
Can anyone else answer this question? This is a quite critical topic for our customers. Thank you.
Hello
Firefox uses a client-side translation engine and performs the translation entirely on your local machine.
When you click "Translate" the browser checks if you have the necessary language model. If you don’t, it fetches the model from Mozilla’s servers. Once that file is on your computer, the actual act of translating happens in your RAM, not on the web.
Will the words from the website be collected or "downloaded"? NO.
Your website content is never "downloaded" to Mozilla, nor is it sent to them to be translated.
The "dictionary" and "grammar rules" (the language model) come from Mozilla to you.
The text of the website you are reading, it never leaves your device.
Firefox is automatically downloading the necessary language files in the background the moment he asks for a translation. It’s a "just-in-time" download, but the translation itself remains private and local.