[Pog42] Sewayaki Kitsune no Senko-san - 12 (BD 1080p x265 FLAC)[BobP].mkv

Any comments on the translation?

Disclaimer: I have only ever released one full-length fansub translation. This kind of review would benefit from the input of a more seasoned fansubber than myself, but I'll offer what help I can out of shared pain (I haven't received much constructive feedback either). I will only be reviewing the "Super Senko-san Time" portion at the end.

(1/3)

Translation:

• Nitpick: おまけの is not translated in the first sign. This indicates that this is an extra scene.
• The second sign is not translated at all. 祝！最終回 means "congratulations! final episode". This seems relevant to include.
• Nitpick: お帰り can mean "welcome home", but in this context it's more like "welcome back", since they don't appear to be at home. Scratch this one. The last shot confirms they are at home.
• "We are here for the company party, everyone has worked on." Aside from missing some details, like the fact that they're celebrating the completion of a project (likely referring to the show itself), this does not sound like idiomatic English. Perhaps "We all helped out with the preparations for today's closing party."
• "But, because it happened..." This is completely wrong. I would have put "You're late! Way to keep me waiting...".
• "I am excited to say to everyone, "Thank you"." Missing a bunch of details. Here's the Japanese sentence for reference: ここまで、　世話やきキツネの仙狐さんは　たたかく応援してくれた皆様、本当にありがとうなのじゃん。A quick and dirty translation would be "To everyone who rooted for us along the way, The Helpful Fox Senko-san would like to thank you from the bottom of my heart."
• "This Shiro-sama dedicated herself to the show". I can't make out the Japanese well enough to comment.
• Nitpick: "Shiro-chan, you always did your best with games." The word 確かに is not translated.
• "We crazily did our best and more." Not idiomatic English, and also missing the detail about deployment tasks, but there are some verbal ticks peppered in that I'm not too familiar with, so I can't offer an alternative.
• Same with Senko's next sentence: too many verbal ticks for me to comment.
• "Our secret episode" Completely wrong. Japanese is その日まで, which pretty straightforwardly translates to "until that day", though a more idiomatic translation might be "until we meet again".
• Nitpick: lines that are basically just mouth sounds don't really need translation.

(2/3)

Timing:

• Keyframe timing is a bit off on the first sign.
• Good job having no gaps between some of the adjacent lines, but your gap threshold was arguably too low (meaning more of the lines should be adjacent than currently are).
• Most lines come in too late. I haven't demuxed the subs so I can't give an exact figure, but feels like a couple hundred milliseconds at least.

Typesetting:

• On the first sign: the colors, font, size, and shadow are all good. I would put the word "Super" somewhere else or make it smaller; in its current placement, it covers up the Japanese text. Nitpick would be to match the way the letters grow and shrink with some karaoke effects, but this isn't strictly necessary.
• On the second sign (if you decide to translate it): you'll need some frame by frame clipping to make it so the text never appears in front of the confetti.
• Main dialogue style: I would prefer less outline personally, but it's passable. I assume you used what was already there for the rest of the episode.

I hope this helps. Sorry I focused so much on negative criticism. If there are more experienced fansubbers out there reading this, please don't hesitate to weigh in, as I probably missed some things.

(3/3)

Thanks Typesetter. First sign comes straight from the other "Super Senko Time" in Episode 10. I just adjusted it to fit the start time the way Episode 12 was stitched together. I did adjust some errors I saw in the timing which I saw in the original Episode 10. Aegisub would not play the video but was required for the .ass enhancements so I had to use Subtitle Edit which would load the video as reference for timing. So it was not the best situation and I have not been dealing much with timing before this because I have mainly been converting .vob to .ass with SE's OCR for ripping old DVDs. As far as translation, I am still learning Japanese and depend too much on dictionaries and translators so I can miss context of words with multiple meanings. Which tools do you use? As far as your nitpick, I was following the conventions the show used because I had just finished rewatching the show. Would you mind if I released a V2 based on your observations? I would have wished that the original releasers or someone else had subbed this but no one was stepping up and it bugged me.

I don't mind at all if you release a v2.

For timing, Aegisub sometimes generates the wrong keyframes from the video file. I use a modified version of this script to generate the correct ones. You don't actually have to load a video file to use keyframes. You can just load the audio, then click "open keyframes" from the video menu and choose the keyframes file you created with that script. Don't use any program other than Aegisub for timing, as it has the best tools along with programmable keybindings to vastly increase efficiency.

As far as translation, the errors you made indicate that your biggest issue is hearing all of the words in a sentence. I would recommend slowing down the audio and really listening to make sure you hear every syllable (in mpv you can slow down / speed up the audio by pressing [ and ] respectively), although this will only get you so far. There's really no replacement for taking the time to learn Japanese grammar, which isn't something you'll get from a dictionary or online translation tool, although if you want my recommendation, Jisho is the best dictionary I've found, and Deepl certainly does a lot better than Google Translate, but it always translates androgynous pronouns as male, so watch out for that. Another piece of general advice is to stop using romaji and learn to read/type katakana and hiragana. This is a walk in the park compared to that next daunting task, learning the kanji, but there are some free online courses and study apps that are pretty good at drilling those. Finally, when you have translated the Japanese into an English sentence, read it aloud to make sure it sounds like something an English speaker would say and change accordingly if it doesn't.

Can you please link me to the tools you use for OCR'ing subtitles? I haven't found any that work for me.

Thanks, I will look back over your message and see if I can do it.

Ahh, keyframes. Since Aegisub would not load the video, it was outputting .ass files that I had to mux using MKVToolNix GUI so part of the problem may be there. All I could do was use best guess timing seen from Subtitle Edit (only used to play .mkv to see timing stopwatch) and adjust by ear/eye.

Grammar and particles are my weak point right now. I was going to take a college level Japanese Language course but it was cancelled due to losing the instructor. I mostly have the kana down. But kanji is a real problem. I lived in Taiwan as a child (my father was in the Army) but I only learned a little spoken Chinese. I was hoping to be further along in the language before I tackled translation but this was a case of it was a small job that bugged me just like the many untranslated intro and special shorts we see. So I gave it a try.

I use Subtitle Edit for VOBSUB OCR. It comes with OCR software loaded in. But it defaults to .srt mode so be sure to change that before saving unless you want that. And I recommend the latest Tesseract OCR method with the Tesseract +LSTM Engine mode with Italic and Music symbol checked and fallback unchecked when doing VOBSUB (OCR window automatically comes up). I do not remember the defaults and have not had much luck with other settings but I am sure they have their place. And I then use MKVToolnix GUI to mux. Do you use something different for that?

Version 2 is out now. Grab that one instead.

I use ffmpeg for muxing.