DeepL in the fall he announced some important news. It can translate voice in real time, expanded the support of typing assistance to other languages and strengthened by a fast Nvidia server on which it can develop new technologies.
The German company started offering two new products Voice for Meetings and Voice for Conversations to companies. In both cases, it is a linguistic AI that can instantly translate spoken word using subtitles. In the first case, it is aimed at conference calls, in the second, at voice calls on mobile phones. The idea is for each speaker to speak their own language while the software takes care of the understanding.
„DeepL is already a leader in text translation, but real-time speech translation is something else entirely. When translating ongoing spoken speech, you encounter incomplete input, pronunciation complications, latency, and the like, which can lead to inaccurate translations and negatively impact user friendliness,” says Jarek Kutylowski, founder and director of DeepLu.
Voice is currently only available in English, German, Japanese, Korean, Swedish, Dutch, French, Turkish, Polish, Portuguese, Russian, Spanish and Italian, but will eventually expand to all languages supported by DeepLem, including Arabic recently . However, all languages already work on the output side, so spoken Spanish can be translated into text Czech, comprende?
Another novelty is related to the voice, it is possible to dictate the text on computers. Until now, this function was only available on mobile phones, where it has a greater justification. When communicating with a stranger, you will have a smartphone at hand rather than a computer.
Who uses Google for correcting and editing texts with the Write tool, it can now also work with Italian and Portuguese. Until now, DeepL only supported English, German, French and Spanish. However, this is where ChatGPT and similar AI will help you just as well, if not better, today.
But just so that DeepL doesn’t lose its breath on universal large language models powering omnipotent chatbots, the company from Nvidia bought DGX SuperPOD. It contains 36 GB200 computers, each consisting of two powerful B200 accelerators, 72 Arm Neoverse V2 cores and 864 GB of memory. Nvidia is to deliver SuperPOD in the first half of 2025, and DeepL wants to train new generations of language AI on it.
DeepL’s New Features: A Comedic Take on Real-Time Translation
Well, folks, gather round because we are diving into some pretty exciting news from a little company called DeepL, which apparently decided that translating is not just for the written word anymore. You know, because why stop at text when you can try to translate an awkward family dinner in real-time? I mean, have you heard a German relative try to order spaghetti in Italian? It’s like a butcher trying to serenade a cow!
So, what’s on the menu this fall, you ask? They’ve introduced Voice for Meetings and Voice for Conversations. Yes, that’s right; now you can bungle your international calls in multiple languages rather than just one. Imagine the fun at your next conference call when everyone is speaking their own language and DeepL is there, trying to keep up like it’s a poor intern at a fast-paced news agency. Talk about pressure!
Now, let’s hear from Jarek Kutylowski, founder of DeepL. He says, “Real-time speech translation is something else entirely.”
Currently, Voice only works in a handful of languages. It’s like a well-meaning tourist who knows how to say “hello” and “thank you” in a dozen languages but gets completely lost when asked for directions. English, German, Japanese, Korean—all available! But if you need a fluent translation into Urdu or Swahili, you’re going to be waiting a bit, my friend—a bit like waiting for your late British friend who promised he’d be on time, but just… isn’t.
Oh, and let’s not forget the new dictation feature. Now you can talk at your computer like a dignified human instead of typing like it’s a race against time. Until recently, this feature was like the exclusive VIP lounge—you had to be on a mobile device for access! But now, it’s open to your favorite snazzy keyboard too. Because who doesn’t want to look like they’re having a heated argument with their laptop?
And wait… DeepL is extending its text correction part of the business to Italian and Portuguese! So if you’re out there using Google to polish your prose, be aware that now there’s competition. But let’s face it; if you’re putting your fate in the hands of ChatGPT and similar AIs, bless your heart. It’s like sending a toddler to do your taxes; you’re asking for trouble.
But how does DeepL intend to compete with those omnipotent chatbots? Why, by beefing up their tech stack with a little help from Nvidia, of course! They’re getting themselves a DGX SuperPOD, which sounds like a futuristic spaceship straight out of a sci-fi movie! This thing is designed to supercharge their language AI—36 GB200 computers packed into one mega-machine! So they’re ready to tackle the toughest language challenges right out of a sci-fi novel.
To sum it up, DeepL is definitely stepping up its game. Before we know it, we might have intergalactic conversations without even leaving our living rooms. But, until then, we’ll have to navigate our way through mishaps, buddying up with AI that stands about as much chance of accurately translating that family feast as you do of winning an argument with your mum about why you should have prawn crackers.
In closing, let’s raise a glass (or a smartphone) to DeepL, and to all the hilarities that real-time translation will undoubtedly bring. Just remember, the next time you’re on a call and hear some bizarre translations: it’s not just tech finding its feet; it’s the world learning to laugh together—one huge communication faux pas after another!
In a significant announcement made by DeepL this fall, the company revealed several groundbreaking developments designed to enhance user experience and broaden functionality. Among these advancements is the capability for real-time voice translation, bolstered by an expanded typing assistance feature that now supports a multitude of languages. This leap forward is made possible by the integration of a robust Nvidia server, which facilitates the development of innovative new technologies.
The renowned German corporation has introduced two novel products: Voice for Meetings and Voice for Conversations, specifically tailored for business applications. These offerings leverage cutting-edge linguistic AI technology to instantly translate spoken words into subtitles. Voice for Meetings is optimized for conference calls, while Voice for Conversations caters to mobile voice calls. This revolutionary approach allows each participant to communicate in their native language with the software seamlessly bridging any language barriers.
„DeepL is already a leader in text translation, but real-time speech translation is a completely different challenge. When translating spontaneous spoken dialogue, complications arise from incomplete input, variations in pronunciation, and latency, which can compromise translation accuracy and user experience,” explained Jarek Kutylowski, founder and director of DeepL. This insight underscores the complexity of achieving effective real-time translation amidst various conversational dynamics.
Currently, the Voice feature supports translations in English, German, Japanese, Korean, Swedish, Dutch, French, Turkish, Polish, Portuguese, Russian, Spanish, and Italian. However, the company anticipates expanding this functionality to encompass all languages available through DeepL, including recently added Arabic. Notably, all target languages can currently have spoken input translated into text, exemplifying the system’s versatility; for instance, spoken Spanish can be translated into Czech text, fostering enhanced communication across diverse linguistic backgrounds.
Another noteworthy enhancement enables users to dictate text directly on computers, a feature that was previously limited to mobile devices. This upgrade significantly enhances user convenience, particularly in situations where one would typically interact with a stranger, making a computer’s dictation feature more accessible and useful than before.
For users who rely on Google for text correction and editing with the Write tool, there is now additional support for Italian and Portuguese. Previously, DeepL’s editing capabilities were restricted to English, German, French, and Spanish. However, the advancement of tools like ChatGPT and similar AI technologies raises the competition, offering equally effective, if not superior, editing services in the current landscape.
To maintain its competitive edge against universal large language models that power omnipotent chatbots, DeepL has made a significant investment by acquiring the Nvidia DGX SuperPOD system. Composed of 36 GB200 computers, each equipped with dual B200 accelerators, 72 Arm Neoverse V2 cores, and 864 GB memory, this powerful hardware will enable DeepL to train new generations of language AI. Nvidia is scheduled to deliver the SuperPOD in the first half of 2025, paving the way for future advancements in AI-driven language technology.
What languages will DeepL support in its voice translation feature?
And Italian. The company has plans to expand this to include all languages supported by DeepL, including Arabic. However, users should note that while input is currently limited to specific languages, output translations are available across the board. For instance, you can vocally express yourself in Spanish, and the software can translate your words into written Czech—talk about a multilingual marvel!
In addition to these voice capabilities, DeepL has introduced a new feature allowing users to dictate text directly on their computers—a function previously only available on mobile devices. This enhancement brings more versatility to the platform, making it easier to communicate effectively, especially in professional settings. After all, who doesn’t prefer to speak their thoughts rather than hammer them out on a keyboard like an over-caffeinated novelist?
Moreover, for those utilizing Google for correcting and editing texts with their “Write” tool, you’ll now find support for Italian and Portuguese as well. Until now, DeepL primarily focused on English, German, French, and Spanish, which puts them in direct competition with other AI-based writing assistants. This indicates a clear shift toward a more comprehensive language support strategy in an increasingly diverse global landscape.
To ensure they keep pace with large language models that drive powerful chatbot applications, DeepL has also secured a DGX SuperPOD from Nvidia—a major investment in cutting-edge technology. This machine is nothing short of a powerhouse, containing 36 GB200 computers with impressive specifications designed to train new generations of language AI. Set to arrive in the first half of 2025, the DGX SuperPOD positions DeepL to tackle even the most complex language challenges ahead.
DeepL is clearly committed to refining its technology and expanding its offerings in voice translation and text editing. As they gear up to challenge established players in the AI language space, the potential for real-time translation to reshape communication is exhilarating. Here’s to a future where language barriers are a thing of the past, and every conversation can be a delightful example of the joys—and occasional hilarities—of global dialogue!