We are frequently asked how it is that DeepL Translator often works better than competing systems from major tech companies. There are several reasons for this. Like most translation systems, DeepL Translator translates texts using artificial neural networks. These networks are trained on many millions of translated texts. However, our researchers have been able to make many improvements to the overall neural network methodology, mainly in four areas. It is well known that most publicly available translation systems are direct modifications of the Transformer architecture. Of course, the neural networks of DeepL also contain parts of this architecture, such as attention mechanisms. However, there are also significant differences in the topology of the networks that lead to an overall significant improvement in translation quality over the public research state of the art. We see these differences in network architecture quality clearly when we internally train and compare our architectures and the best known Transformer architectures on the same data. Most of our direct competitors are major tech companies, which have a history of many years developing web crawlers. They therefore have a distinct advantage in the amount of training data available. We, on the other hand, place great emphasis on the targeted acquisition of special training data that helps our network to achieve higher translation quality. For this purpose, we have developed, among other things, special crawlers that automatically find translations on the internet and assess their quality. In public research, training networks are usually trained using the “supervised learning” method. The network is shown different examples over and over again. The network repeatedly compares its own translations with the translations from the training data. If there are discrepancies, the weights of the network are adjusted accordingly. We also use other techniques from other areas of machine learning when training the neural networks. This also allows us to achieve significant improvements. Meanwhile, we (like our largest competitors) train translation networks with many billions of parameters. These networks are so large that they can only be trained in a distributed fashion on very large dedicated compute clusters. However, in our research we attach great importance to the fact that the parameters of the network are used very efficiently. This is how we have managed to achieve a similar translation quality even with our smaller and faster networks. We can therefore also offer very high translation quality to users of our free service. Of course, we are always on the lookout for very good mathematicians and computer scientists who would like to help drive development, further improve DeepL Translator, and break down language barriers around the world. If you also have experience with mathematics and neural network training, and if it fulfills you to work on a product that is used worldwide for free, then please apply to DeepL! We’re excited to announce glossary support for the DeepL API! Glossaries are available to both DeepL API Free and Pro subscribers. You can create an account here if you don’t already have one. We first added glossary support for our web translator and desktop apps in 2020, and it’s become a much-loved feature amongst our users. The use cases are many and varied—whether you’re translating… ...there’s often brand- and industry-specific vocabulary that you need to account for. Glossaries allow you to specify your own translations for words and phrases, making it possible to customize your translations consistently and at scale so you can deliver the best possible results to your users without pushing a bunch of manual work onto your translation and content teams. And it's an especially important capability for us to offer to our API users, who are often building automated translation workflows that, with the help of glossaries, can be automated even further. Glossary support for the API will also make it possible for CAT tool providers with a DeepL plug-in to build glossary functionality into their products. If you’d like to learn how to use glossaries in the API, or you’re ready to get started, you can check out the API documentation. In the rest of this post, we’ll share a bit more about what you can expect from the feature. Currently, glossaries for the API can be created for the same language combinations available for glossaries on the DeepL web translator and desktop apps: Our team is working on adding more language combinations soon. Glossaries created via the DeepL API are distinct from glossaries created via the DeepL web translator and desktop apps. This means that API glossaries can't be used on the web translator and desktop apps and vice versa. And both DeepL API Free and Pro users can create up to 1000 glossaries. The maximum size limit for a glossary is 10 MB. To wrap it up, here’s everything you’ll need to get started with glossaries in the API and to get help if you need it: We’re excited to announce the release of our Python client library for the DeepL API. This is the first programming language-specific library we’ve built for the API, and our goal is to make it much easier for developers working with Python to build applications with DeepL. Just how much easier? Here's a text translation example. What used to be: Is now: We intend to support all functions of the DeepL API with the Python library, though support for new features may be added to the Python library after they're added to the API. If that’s all you needed to hear, and you’d like to get started right away, you can find the Python client library and documentation here. (By the way: you'll need an API authentication key to use the Python library. If you don't yet have a key, you can sign up for a Free or Pro API plan here.) In the rest of this post, we’ll share more about the Python library project and how you can get involved. Python was an ideal fit for our first client library. It's one of the most widely used programming languages in the world, and it’s popular amongst our API user base. It’s also a language that we have a lot of experience with at DeepL. That said, use cases for the DeepL API are wide ranging, and there are many languages beyond Python that are important to our users. We’re exploring other client libraries we might develop in the future, and if you have any feedback on the languages that are most important to you, please get in touch and let us know. The Python client library is also DeepL’s first ever open source project. We’re always looking for new ways to learn from our users and ensure that our products are solving real problems for them. We believe that open sourcing the project is the best way to foster transparent discussion about what users need and what problems they’re running into, and we encourage you to create an issue if you have ideas or feedback. Of course, GitHub issues isn’t the right place for every challenge you might run into while working with the Python library. If your help request includes sensitive data, please contact our support team. We’re happy to help. To recap, if you’d like to get started with DeepL’s Python client library, you can: Thanks for reading. We’re excited to see what you build. Have you ever wished for instant fluency in another language? Maybe you were traveling and couldn’t read a menu, or you met a new friend but didn’t speak the same language. With our new mobile translation app for iOS, you can say goodbye to language barriers. No matter where you are or who you’re with, you can get your iOS device out and easily access DeepL’s accurate translations. The app is accessible on iPhone and iPad and allows users to translate on the go. It supports all 26 DeepL languages and language variants, as well as other DeepL benefits like fast translations, quick language detection, and best-in-class translation quality. Additionally, you can try the speech to text translation feature, which allows you to use your device’s microphone to translate speech input. It also works the other way around with the text to speech feature, which allows you to listen to your text input and its translation. This is yet another exciting milestone for our team at DeepL as we continue to work towards eliminating language barriers and bringing cultures together. Eager to start translating on your iOS device? You can download the app for free here. We're excited to share that DeepL Translator now supports thirteen new European languages: Bulgarian, Czech, Danish, Estonian, Finnish, Greek, Hungarian, Latvian, Lithuanian, Romanian, Slovak, Slovenian, and Swedish. Over the years, many people have asked us to add more languages to our translator so that they could enjoy DeepL’s accurate translations. We’re thrilled to be able to honor several of those requests at once with the release of the new languages. With this launch, DeepL can now reach an estimated 105 million more native speakers around the world with natural-sounding, high-quality translations. This allows us to further our company vision of breaking down language barriers and bringing cultures closer together. We’re continually working to ensure the translation quality for all languages remains exceptional and that the translation process is seamless for all our users. And if your preferred language isn’t offered yet, don’t worry, we will add more languages in the future. From “Sziasztok” to “Hej”, we’re ready to say hello to our new users and are already looking forward to their feedback. All thirteen languages are now available on DeepL.com, via the DeepL API, and in our desktop apps, DeepL for Windows and DeepL for Mac. Ready to try out the new languages for yourself? Head over to DeepL Translator now to start translating. DeepL has now made it easier to get English translations in your preferred style: DeepL Translator can now produce translations that reflect the particularities of American and British English. Variations in spelling are covered: American neighbors become British neighbours, and they can call each other on their cell phones or on their mobile phones, depending on your preference. Similarly, terms that have completely different equivalents across the pond are also taken into account. American cookies become British biscuits. That is, of course, unless they’re on your web browser. This expanded control over DeepL Translator enables you to more quickly get translations that respect the language conventions that you prefer. To set the language variety you would like, simply select either “English (American)” or “English (British)” as the target language. American and British English are now available as target languages on DeepL.com, via the DeepL API, and in our desktop applications, DeepL for Windows and DeepL for Mac. Combining this feature with other features, such as the Glossary, allows you to take full control of the English texts DeepL Translator produces for you. *Please note that these two language variants are not currently available for translations from Japanese and Chinese. These two source languages will be added in the near future. Since we added Japanese to DeepL Translator, we’ve been amazed by how quickly people in Japan have taken to using our services. We made it a priority to make our paid subscription service, DeepL Pro, available in Japan and we can now announce that that has become a reality. All DeepL Pro subscription plans, providing added data protection, unlimited translation capacity, API access, and more, are now available in Japan. Japanese customers also benefit from prices in Yen. We continue to expand DeepL Pro coverage around the world and will add further countries and currencies in the near future. What’s better than finding the exact right word to express yourself? Always finding the exact right word to express yourself! Our new Glossary feature gives you the power to determine exactly how DeepL Translator should translate terms for you. Whether you translate legal contracts, technical manuals, or company newsletters, the Glossary can save the terms you use and the translations you like, ensuring consistency and saving you time. All you need to do is click on a word in the translated text, click on your preferred formulation, and save the Glossary pair when asked. Alternately, you can click on “Customization” and enter it manually. You can create Glossary entries for nouns, verbs, adjectives, adverbs, and even multi-word entries, and DeepL Translator will adapt the grammar and formulation to accommodate your preferences. You always have the option to turn the Customization feature on or off, giving you full control over how and when the rules you create affect DeepL Translator. Free users of DeepL Translator can create a limited number of Glossary pairs, whereas DeepL Pro subscribers can create as many as they like! Currently, Glossary pairs can be created for the following language combinations: German into English, French into English, English into German, and English into French. Further language combinations will be added in the near future, as will further options to customize DeepL Translator’s results and make the translations your own. We are delighted to announce that DeepL Pro is now available in the Land of the Free and the True North! This has been a longstanding goal of ours and we are very proud to welcome our American and Canadian friends. All DeepL Pro subscription plans, providing added data protection, unlimited translation capacity, API access, and more, are available immediately. Additionally, we are very happy to allow our North-America-based clients to pay for their DeepL Pro subscriptions in their local currencies, US and Canadian dollars. We’re constantly working to enter new markets and will be making DeepL Pro available in further countries in the near future. Please note that while businesses in Quebec may sign up as of today, registration will be open for private citizens shortly. Que legal! DeepL Translator is now also capable of translating into Brazilian Portuguese! The Portuguese language, in all its varieties, is spoken all around the world by millions of people on several continents. The country with the most Lusophones is by far Brazil, with over 200 million people. To reflect some of this linguistic diversity, our researchers have taught our algorithms to also produce translations in the variety of Portuguese used in Brazil. This allows DeepL Translator to capture the nuances and unique vocabulary of Brazil, as well as the language of millions more in Portugal, Angola, Mozambique, and elsewhere. Some variations in the Portuguese language are clear to see in the translations of the phrase “The girl is having breakfast.” Portuguese: A rapariga está a tomar o pequeno-almoço.Brazilian Portuguese: A menina está tomando o café da manhã. The terms for “the girl” (a rapariga/a menina) and “breakfast” (o pequeno-almoço/o café da manhã) reflect regional variations, as does the form of the verb tomar; Brazilian Portuguese prefers the gerund, whereas a construction using the infinitive is more common in Portugal and other countries. The two variations of Portuguese are now available on DeepL.com. To select the version you prefer, simply click on the language next to “Translate into” and select either “Portuguese” or “Portuguese (Brazilian).” Enjoy the many delights of the Portuguese-speaking world!