Hollywood Localizer Deepdub Launches Enterprise Generative AI Dubbing and Translation Platform Deepdub Go
Audiovisual localizer for film and TV Deepdub is rolling out a generative AI-powered platform for enterprises to translate and dub videos. The new Deepdub Go platform leverages proprietary generative AI models to automate the normally time-intensive dubbing process while maintaining a high-quality audio track suitable for in speed and scale across industries beyond entertainment.
Deepdub is known for working with Hollywood studios to dub dozens of movies and shows. The company usually handles the entire process, but Deepdub Go makes it possible for clients to handle much of it on their own. Users simply upload a video they want to dub, then select the languages and emotional style they want to hear, and Deepdub’s AI handles transcription, translation, voice cloning, and mixing to output a final localized version. Deepdub believes this is far more versatile than relying on automated subtitles. The company envisions Deepdub Go as being useful for sectors like corporate training, academia, and marketing. Deepdub Go’s self-serve accessibility aims to make the service attractive to any creator and is available through a web interface and an API that supports integration with existing media workflows.
“We’ve always been using generative AI models, but now we’ve encapsulated and built a full platform around it. We can [design] models from text or video and audio. What we deliver to studios has services on top of that because they don’t want to work on a platform; they want an end-to-end solution,” Deepdub CEO Ofir Krakowski told Voicebot in an interview. “Deepdub Go took the features most important to people and packaged them together and made them more automatic and simple to use. The audio is separated, splitting dialogue from music and effects, then glued back together after the dub is made.”
Deepdub Go’s large language models (LLMs) bring natural voice performances to the dubbed audio, including emotions. Creators can guide the desired tone and delivery through text prompts or voice recordings. The platform also enables editing output voice characteristics, languages, timing, and more for professional polish. Users can choose from over 65 language options. The platform can translate the voice on a video or employ an existing voice model from its catalog.
“We aim for theatrical grade technology, like a movie theater or feature film, so we needed to not just create text-to-speech [tools] but also expressive voices,” Krakowski said. “The next challenge is going real-time. Live content has its own challenges but I believe we can provide live dubbing to sports and live events.”