Language
Country/Area
No result found
In-depth understanding of DeepL's powerful algorithm: Why CNN can make translation more natural?
As globalization accelerates, the need for translation is increasing. As a neural machine translation service, DeepL Translator was launched in August 2017 and quickly occupied a place in the market. The translation system, owned by Cologne-based DeepL SE, has expanded from initially offering translations into seven European languages to supporting 33 languages since its initial launch. DeepL's translation algorithm relies on convolutional neural network (CNN) technology, which makes its translation results more natural.
DeepL's translation quality is considered to be more natural than other competing products, mainly due to the convolutional neural network algorithm behind it.
Translation method
DeepL translator uses a unique algorithm based on the structure of convolutional neural networks, which are trained on the Linguee database. According to the developers, this new neural network architecture is able to produce more natural-sounding translations that often outperform similar services. The supercomputer used by DeepL can reach 5.1 petaflops of computing power and runs in Iceland, which uses hydroelectric power.
CNNs are particularly well-suited to processing long, coherent sequences of words, which is why other competing services haven't made widespread use of this technology.
Although DeepL still has limitations in some aspects, its shortcomings have been compensated by some known assistive technologies, which makes it excellent in the naturalness of translation.
Free and subscription services
Users can use the DeepL translator for free, but each translation is limited to 1,500 characters. In addition, the platform can also translate Microsoft Word and PowerPoint files as well as PDF files up to 5MB in size. For users who require more functionality, DeepL also offers a paid subscription service, DeepL Pro, available since March 2018, which includes API access and plug-ins for computer-assisted translation tools.
Unlike the free version, DeepL Pro's translated text is not saved by the server, and the maximum character limit is removed. Users can calculate the monthly fee based on the number of characters required.
Supported languages
As of December 2024, DeepL translation service has supported translation in up to 33 languages, fully meeting the needs of users.
Historical evolution
The DeepL translation system was developed internally at Linguee in 2016 by a team led by Chief Technology Officer Jarosław Kutyłowski. On August 28, 2017, DeepL translator was officially launched, initially supporting translation into English, German, French, Spanish, Italian, Polish and Dutch. At the time of its launch, DeepL claimed to have surpassed all competitors, including Google Translate, Amazon Translate and Microsoft Translate, in blind tests and BLEU scores. With the launch of DeepL, Linguee's company name was changed to DeepL GmbH and funded through advertising on its sister website linguee.com.
Over time, DeepL's language support scope has gradually expanded, and new languages have been added one after another, including Brazilian Portuguese, Russian, Simplified Chinese and Japanese. The launch of these languages has gradually made the DeepL translator gain praise in the industry. .
DeepL Write
In November 2022, DeepL launched a tool called DeepL Write to improve the quality of monolingual texts in English and German, and the subsequent public beta version was released on January 17, 2023. In the summer of 2024, DeepL announced the addition of support for French and Spanish to further expand its text writing assistance capabilities.
Market response
Overall, the DeepL translator has received widespread praise. Many media outlets affirmed its accuracy and pointed out that its translation was more precise and detailed. Although some reviews mention that the number of languages DeepL can translate is relatively small, its presence in the market cannot be underestimated. DeepL has also won multiple awards for its outstanding technical achievements, including the 2020 Webby Award.
Whether DeepL can continue to maintain its advantages in the upcoming intelligent translation competition is still the focus of the industry.
DeepL's success is not only reflected in the improvement of its translation quality, but also in its desire to continuously introduce new products to meet user needs. However, in this rapidly changing technology field, can DeepL continue to lead the trend and maintain its competitiveness?
