DeepL launches DeepL Voice, Making Real-Time Translation Easy

DeepL has released a major update with DeepL Voice—a tool aimed at making language barriers a thing of the past. Known for its accurate text translations, DeepL is now taking a step further by allowing users to communicate with colleagues, clients, or anyone in real time, no matter what language they speak.

What DeepL Voice does

With DeepL Voice, there’s no need to brush up on another language before a meeting. Simply speak in your language, and DeepL translates it into text captions in real time. Here’s what stands out:

• 13 supported languages for audio input: including English, Japanese, German, and Korean.
• 33 languages for text captions: allowing smooth conversation flow, whether you're speaking Dutch, Russian, or beyond.

And perhaps most importantly, DeepL Voice translates while you talk. This means it’s ready to support virtual boardrooms, quick chats, or even casual coffee conversations with clients.

Two versions for different needs

Recognizing that communication needs vary, DeepL offers two versions of DeepL Voice:

• DeepL Voice for meetings: Designed for virtual settings where each participant can speak their native language while captions translate for others. Currently, it’s compatible with Microsoft Teams, covering much of DeepL’s B2B client base.
• DeepL Voice for conversations: Optimized for in-person interactions on mobile, making it easy for frontline employees to engage with customers or colleagues, regardless of language.

Whether collaborating globally or speaking face-to-face, DeepL Voice adapts to the setting, streamlining communication.

The challenges DeepL had to solve

Real-time translation sounds simple but is complex in practice. DeepL Voice faced unique challenges such as incomplete sentences, accents, and the speed of live conversation. To overcome these, DeepL took a different route from many other tools, building its own models specifically for translation rather than using generic language models. This tailored approach allows DeepL Voice to deliver fast, accurate translations ideal for live, real-time interactions.

DeepL’s approach to privacy

In today’s data-sensitive environment, security is essential. DeepL assures users that DeepL Voice processes translations on its servers without storing any data for model training. Fully GDPR-compliant, DeepL Voice prioritizes user privacy, ensuring conversations remain private.

What’s next for DeepL Voice?

This release is just the beginning. Currently, DeepL Voice focuses on text-based translations in real time, but CEO Jarek Kutylowski has hinted at a broader vision that includes instant audio and video translations. For now, DeepL Voice is set to help businesses with the real-time, text-based captions they need to connect, collaborate, and communicate without language holding them back.

Source: https://www.deepl.com/en/blog/deepl-voice