3 ms·

The problem with Google is its translation quality. Not sure about Korean, but Japanese/English (either way) definitely isn't there. For Japanese to English, t

The problem with Google is its translation quality. Not sure about Korean, but Japanese/English (either way) definitely isn't there. For Japanese to English, the transcription alone is already pretty inaccurate (usable if you know some Japanese; but then again you already know Japanese!)

sebzim4500 2y ago
Interesting, I was in Japan a few months ago and I found google translate to be pretty good. Even when Hotels etc. provided information in English I found it was better to use google lens on the Japanese information. I can't say much about the quality of English -> Japanese translation, except that people were generally able to understand whatever came out of it.
- thrdbndndn 2y ago
  It's usable as a tool for quick communication or reading instructional text. But don't expect to be able to use it to read actual literature or, back to the topic, subtitling a TV series or a YouTube video without misunderstanding.
mathgorges 2y ago
This hasn't been my experience with English/Japanese translation with Google Translate. For context I used Google Translate for pair programming with Japanese clients 40 hours per week for about 6 months, until I ponied up for a DeepL subscription. As long as you're expressive enough in English, and reverse the translation direction every now and again to double check the output then it works fine.
- thrdbndndn 2y ago
  As I mentioned in another reply, the scenario here is translating "artistic" or "real-world" (for lack of a better term) literature accurately—whether it's a novel, a YouTuber's video, casual conversation, or blog posts/tweets with internet slang and abbreviations. In these cases, getting things 95% right isn’t enough to capture the nuances, especially when the author didn’t create the content with translation in mind (which I believe matches your experience). Machine translation for instructional or work-related texts has been "usable" for years, way before LLM emerged. LLM-based translation has certainly made significant progress in these scenarios—GPT-4, for example, is fully capable IMHO. However, it's still not quite fast enough for real-time use, and the smaller models that can run offline still don't deliver the needed quality.
- numpad0 2y ago
  English -> Japanese machine translations(whether it's GT or DL or GPT) are fairly usable these days in the sense that it reduces interpretation workload to trivial amount especially with typical skillset of a Japanese white collar workers, or not perfect in the sense that the fact that the output is a translation is always apparent to native speakers - but that is the case too even with offline human translators, so could be a moot point. Anyway, the current state of affairs float somewhere comfortably above "broken clock" and unfortunately below "Babelfish achieved", so opinions may vary.
maeil 2y ago
The leading LLMs are already very strong at translation (including EN<>JP, Korean is more model-dependent), so then what you want is simply Google Translate but powered by a strong LLM? I'm sure there's already dozens of such wrappers that offer just that.

sebzim4500 2y ago

Interesting, I was in Japan a few months ago and I found google translate to be pretty good. Even when Hotels etc. provided information in English I found it was better to use google lens on the Japanese information. I can't say much about the quality of English -> Japanese translation, except that people were generally able to understand whatever came out of it.

thrdbndndn 2y ago
It's usable as a tool for quick communication or reading instructional text. But don't expect to be able to use it to read actual literature or, back to the topic, subtitling a TV series or a YouTube video without misunderstanding.

mathgorges 2y ago

This hasn't been my experience with English/Japanese translation with Google Translate. For context I used Google Translate for pair programming with Japanese clients 40 hours per week for about 6 months, until I ponied up for a DeepL subscription. As long as you're expressive enough in English, and reverse the translation direction every now and again to double check the output then it works fine.

thrdbndndn 2y ago
As I mentioned in another reply, the scenario here is translating "artistic" or "real-world" (for lack of a better term) literature accurately—whether it's a novel, a YouTuber's video, casual conversation, or blog posts/tweets with internet slang and abbreviations. In these cases, getting things 95% right isn’t enough to capture the nuances, especially when the author didn’t create the content with translation in mind (which I believe matches your experience). Machine translation for instructional or work-related texts has been "usable" for years, way before LLM emerged. LLM-based translation has certainly made significant progress in these scenarios—GPT-4, for example, is fully capable IMHO. However, it's still not quite fast enough for real-time use, and the smaller models that can run offline still don't deliver the needed quality.
numpad0 2y ago
English -> Japanese machine translations(whether it's GT or DL or GPT) are fairly usable these days in the sense that it reduces interpretation workload to trivial amount especially with typical skillset of a Japanese white collar workers, or not perfect in the sense that the fact that the output is a translation is always apparent to native speakers - but that is the case too even with offline human translators, so could be a moot point. Anyway, the current state of affairs float somewhere comfortably above "broken clock" and unfortunately below "Babelfish achieved", so opinions may vary.

maeil 2y ago

The leading LLMs are already very strong at translation (including EN<>JP, Korean is more model-dependent), so then what you want is simply Google Translate but powered by a strong LLM? I'm sure there's already dozens of such wrappers that offer just that.