Q: Translation word count for multibyte languages?
Hey Kunal, sorry but here's another question:
How do your credits work when translating multibyte (or double-byte) encoded languages — such as Japanese, Chinese and Korean — compared to English word translations?
My understanding is a single English alphabet letter is 1 byte, while a Japanese character, for example, is 2 (or sometimes 3) bytes. In terms of word structure, I believe Japanese word count is typically estimated by the number of characters, not individual words (like English), and this count is often less than half the number of English words in a translation.
Does each Japanese character count as 1 credit? Or is there some kind of 'weighting' system that compensates credit usage based on the language?
This is important because it may drastically alter how many credits we need to consider for our non-English sites.
Kunal_MultiLipi
Apr 14, 2025A: Great question — and you’re absolutely right to bring this up, especially when working with multibyte languages like Japanese, Chinese, or Korean (CJK languages). Here’s how it works on MultiLipi:
👉 We don’t count characters or bytes — instead, we calculate based on the English source word count, no matter what the target language is.
So even if your translated content ends up being shorter in Japanese or Chinese (because of compact characters), the credits consumed are based entirely on the original English content — not the byte size or target language character count.
✅ Example:
If your English page has 1,000 words and you translate it into Japanese, it will still consume 1,000 words from your translation credit balance — even if the Japanese version only uses 600 characters.
This way, there’s no extra “weighting” or penalty for translating into multibyte languages. So feel free to go global — your credits are used fairly across all languages 🌏
Let me know if you’d like help estimating usage for your site!
iXzenoSPLUS
Verified purchaser
Wow, didn't expect an answer so quickly! That's super impressive and many thanks for clarifying on the other follow-up questions :)
I understand your explanation, however what about when it's the other way around and the source language is not English?
For example, what if we have a website created in Japanese, and we'd like MultiLipi to translate that to English and other languages?
Glad it helped — and great follow-up!
If your source language is Japanese, we still count words based on the source, not the English output. Since Japanese doesn’t use spaces, we use smart tokenization to estimate the word count fairly (via Google/Microsoft models).
So don’t worry — you won’t be overcharged for shorter or multibyte text. It’s all balanced across languages.
iXzenoSPLUSVerified purchaser
Got it, thanks so much for al your help Kunal! Will evaluate our potential usage and take it from there. Keep up the awesome work!
Thank you so much, really appreciate the kind words! 🙌
We’re here whenever you’re ready — feel free to reach out anytime if you need help or have questions along the way.
Wishing you all the best with your project! 😊