That is, DeepL is a little better at translating the actual meaning and tone of content, rather than just word-for-word translations that might technically be correct, but still sound a bit awkward or lack context. In general, people praise DeepL for having more natural-sounding translations and doing better with local idioms and slang. This is especially true for European language pairs, as we noted above. The article is in Japanese – but you can use one of these services to translate it!įinally, based on general sentiment around the web, most people find DeepL’s translations to be a little bit more accurate and natural-sounding. Gigazine, a popular Japanese blog, also ran some tests of DeepL’s Japanese translations and found them to be more accurate than Google Translate.
It’s only presenting the data for situations where one translation was rated to be better than others by the professional translators.īut in those specific situations (where one set of translations was noticeably better than others, instead of just being equal/identical), DeepL was the service that had the better translations most of the time for the language pairs that they tested: This data does not include situations where the translations were rated equally good (or identical).
The way that DeepL presents this data is a bit tricky, though. They then asked professional translators to do a blind test and rate the translations. However, DeepL generally fares a bit better than Google Translate in blind tests, especially when it comes to European language pairs.įor example, DeepL translated 119 different paragraphs using translations from DeepL, Google, Amazon, and Microsoft. It’s hard to give a single conclusion for accuracy because it depends in some part on the specific language pairs that you’re translating. If you’re using automatic translation, one of your biggest considerations is probably the accuracy of the translations that your chosen service generates.įor that reason, we’re going to start our Google Translate vs DeepL comparison with a look at the accuracy of each service, based on some actual studies and general user opinions.
0 kommentar(er)
