DeepL, known for its text translation tools, has launched a voice-to-voice translation suite today that caters to meetings, mobile and web conversations, and group chats for frontline workers via custom apps. Additionally, the company is offering an API for developers and businesses to create tailored solutions, like for call centers.
“Voice was a natural progression for us after years in text translation,” DeepL CEO Jarek Kutylowski told TechCrunch. He highlighted the absence of a strong real-time voice translation product.
The main challenge is balancing low latency and accuracy in real-time translations. DeepL is introducing add-ons for Zoom and Microsoft Teams, enabling real-time translation audibly or via on-screen text. This is under early access, with organizations invited to a waitlist. DeepL also offers solutions for in-person or remote mobile and web conversations.
Users can join group conversations, such as training sessions via QR code. The voice-to-voice tech also adapts to specific vocabularies, including industry terms and names.
Kutylowski noted AI’s role in reshaping customer service, providing language support where staff is not readily available or costly.
DeepL controls its voice-to-voice stack, currently converting speech to text for translation and back to speech. With years of text translation experience, it aims for a direct voice translation system, bypassing text.
Competitors include Sanas, which raised $65 million to alter call center agents’ accents in real-time, Camb.AI in Dubai focusing on media translation, and Palabra, funded by Alexis Ohanian’s Seven Seven Six, aiming for real-time translation preserving original voice and meaning.