Community Profile
-
Join Community
-
Subscribe
-
Post to Community
-
Track
-
Gift Paid Account
omtehuilen's Journal
Free Account
Created on 2017-12-22 21:36:42 (#3325636), last updated 2018-03-21 (380 weeks ago)
3 comments received
11 Journal Entries, 26 Tags, 0 Memories, 1 Icon Uploaded
Name: | omtehuilen |
---|---|
Membership: | Open |
Posting Access: | Select Members, Moderated |
Google Translations of obscure animanga materials that don't have official or fan translations. Because I have reached the point where anything is better than nothing. Check the community tags for a list of my "projects".
My usual modus operandi is as follows:
I take scans or photographs of pages of Japanese-language novels or comics, run them through OCR software to convert the text in the images into copy-pastable text on a webpage, and then run that text through Google Translate. The weaknesses in this method start at the OCR software. It has a small but not non-existent margin of error when recognizing characters, mostly when used on unclear images. It also doesn't always handle Roman characters in Japanese text, numbers, or punctuation well; oftentimes the website I use flat out refuses to read periods. And it tends to spit short sentences or sentence fragments left alone in their own rows out in random places. I fix all of that to the best of my ability and my patience, but suffice it to say that the raw text is not 100% guaranteed to be flawless to start with. And the less said about Google Translate the better. Beyond character names + obvious him/her pronoun mix-ups and canon terminology, I make only very VERY minor 'cosmetic' edits to Google Translate's results. I understand only the barest amount of Japanese and as such, can't usually tell whether I'm making a garbled sentence better or worse. (Also, I'm lazy.)
Let me repeat that before anybody starts complaining about how much my "translations" suck: I KNOW THEY DO. THEY ARE LITERALLY JUST GOOGLE TRANSLATIONS.
Still. Google Translate is better than NOTHING.
My usual modus operandi is as follows:
I take scans or photographs of pages of Japanese-language novels or comics, run them through OCR software to convert the text in the images into copy-pastable text on a webpage, and then run that text through Google Translate. The weaknesses in this method start at the OCR software. It has a small but not non-existent margin of error when recognizing characters, mostly when used on unclear images. It also doesn't always handle Roman characters in Japanese text, numbers, or punctuation well; oftentimes the website I use flat out refuses to read periods. And it tends to spit short sentences or sentence fragments left alone in their own rows out in random places. I fix all of that to the best of my ability and my patience, but suffice it to say that the raw text is not 100% guaranteed to be flawless to start with. And the less said about Google Translate the better. Beyond character names + obvious him/her pronoun mix-ups and canon terminology, I make only very VERY minor 'cosmetic' edits to Google Translate's results. I understand only the barest amount of Japanese and as such, can't usually tell whether I'm making a garbled sentence better or worse. (Also, I'm lazy.)
Let me repeat that before anybody starts complaining about how much my "translations" suck: I KNOW THEY DO. THEY ARE LITERALLY JUST GOOGLE TRANSLATIONS.
Still. Google Translate is better than NOTHING.
Administrators
Members [View Entries]
To link to this user, copy this code: