DeepL – The free online translator that challenges Google Translate

DeepL – The free online translator that challenges Google Translate

DeepL trains artificial intelligence to understand texts and translate them.
It is now at your disposal.

Are you looking for a FREE and ONLINE translator?

Today I offer you DeepL that offers you free translation quality is unmatched, with a significant decrease in translation errors compared to Google Translate and Microsoft Translate

DeepL continues to develop artificial intelligence products that can expand human possibilities, overcome language barriers and bring cultures closer together.

The European technology company DeepL launches  DeepL Translator , which offers unparalleled quality translations and marks a turning point in the field of neural machine translation systems. From today, it is available online for everyone, free of charge, at  www.DeepL.com .

Despite the volume of money and the amount of resources that major Internet companies have invested in the development of neural networks, DeepL can boast the  most accurate and natural machine translation system in the world . When the user enters a text, DeepL’s artificial intelligence is able to capture even the slightest nuances of meaning and reproduce them in translation, better than any other service.

In the blind tests that compared DeepL Translator with the competition, the translators preferred DeepL’s results with a factor of 3: 1. Automated tests also confirm this. In the field of automatic translation, the reference standard for evaluating the performance of a system is the BLEU score (Bilingual evaluation understudy) which compares the mechanically translated texts with those produced by a translator. DeepL Translator broke the record in BLEU scores. *

“We have made significant improvements in the architecture of neural networks,” explains Gereon Frahling, founder and chief executive officer. “By arranging neurons and their connections differently, we have allowed our networks to map natural language more completely than any other neural network to date.”

DeepL’s revolutionary neural architecture uses the power of a supercomputer in Iceland, with a capacity of 5.1 petaFLOPS (5 100,000,000,000,000 operations per second), enough to translate a million words in less than a second. “Thanks to the abundant availability of renewable energies, we can train our neural networks in Iceland in a cost-efficient way. We will continue to invest in high-performance hardware. ” explains technical director Jaroslaw Kutylowski.

The DeepL team uses this supercomputer to train neural translation networks on a large collection of multilingual texts. During the training, the networks examine a multitude of translations and learn to translate independently, with correct grammar and wording. To this end, DeepL can rely on the success of its first product: Linguee, the largest search engine for translation in the world. Over the past decade, DeepL has collected over a billion high quality translated texts, collecting the best possible training material for a network of.

DeepL Translator currently supports 42 language combinations between Italian, English, German, French, Spanish, Dutch and Polish. Neural networks are already training to master other languages ​​such as Mandarin, Japanese and Russian. DeepL also plans to release an API in the coming months, allowing its higher quality translations to enhance other products such as digital assistants, dictionaries, language learning apps and professional translation programs.

“DeepL’s ambitions are not limited to translation,” says Gereon Frahling. “Our neural networks have developed a level of understanding of the text that opens up new, exciting possibilities.”

* The previous record of the last WMT 2014 set, widely used for English-German, was 28.4, achieved by the Google research department. In the same set, DeepL’s public translation service gets a BLEU score of 31.1. For Sep 2014 for English-French, DeepL Translator gets a BLEU score of 44.7, surpassing the previous record by over 3 points.

The history of DeepL

  • August 2017– The DeepL team launches DeepL Translator. With never-before-seen translation quality, DeepL Translator sets a new standard in neural machine translation. The company changes its name from Linguee to DeepL, with reference to a new focus on deep learning and the development of sophisticated products based on the neural network system.
  • May 2017– DeepL Translator reaches a level of quality never seen before in automatic translation, in all language combinations. At the newstest 2014 ENDE standard translation test, DeepL Translator achieves a BLEU score of 31.1 as the world record – the best previous score being 28.4. Translation quality is unmatched, with a significant decrease in translation errors compared to Google Translate and Microsoft Translate. The whole DeepL team devotes its full attention to the development of DeepL Translator for the public.
  • Linguee exceeds 10 billion page views and continues to grow.
  • 2016– The team starts working on a neural network system capable of translating any type of text. It will be called DeepL Translator. The system is developed starting from many years of experience in the field of machine learning for natural language processing. The neural network system is capable of learning from a billion higher quality translations provided by Linguee’s crawlers.
  • 2014– Linguee develops sophisticated machine learning systems that learn from billions of higher quality translations. The algorithms learn to complete dictionary entries, to provide valid examples, and to recognize linguistic information such as parts of the course, grammatical generals and inflections. Linguee starts recruiting hundreds of professional lexicographers all over the world to evaluate and correct the contents generated by the algorithms.
  • 2010– Just one year after launch, the Linguee service attracts more than one million users every month. Linguee adds various new language combinations as it is about to become the world’s most used online dictionary.
  • 2009– Linguee – the first translation search engine – is launched. For the first time, it is possible to research how a certain expression has been translated by others. The service quickly attracts the attention of professional translators, entrepreneurs, etc.
  • 2007– Gereon Frahling leaves his post at Google Research in New York to start working on a search engine for translations. Together with his friend Leo Fink he worked for 18 months on the development of crawlers and automatic translation systems capable of evaluating the quality of translations.

About DeepL

DeepL is a deep learning company that develops AI systems for languages. The company was founded in 2009 as Linguee and introduced the first Internet search engine for translations. Linguee has responded to over 10 billion queries from over 1 billion users. It ranks among the top 200 most used websites in France, Germany, Spain and other countries.

In 2017, the company introduced DeepL translator, a machine translation system capable of producing translations with an unprecedented quality, available for free on DeepL.com/translator .

DeepL continues to develop artificial intelligence products that can expand human possibilities, overcome language barriers and bring cultures closer togethe

 

Leave a Comment