The YouTube media localization was delivered in two phases, which covered 16 documentary movies. Initially, seven titles and a total of 697 minutes were AI-dubbed from English and French into Brazilian Portuguese, German, and Latin American Spanish (LAS). Next, Dubformer enabled Brazilian Portuguese, Italian, Russian and Castilian Spanish voice-overs, covering a total of 1,267 minutes of factual shows.

“We have been experimenting with AI dubbing tools since the early days and are aware of their limitations for certain types of content,” said Samuel Zaslavsky, CEO of W4tchTV. “In our quest for an end-to-end localization service that seamlessly blends AI with the human touch, we’ve tested various solutions. With Dubformer, we witnessed a significant improvement in the quality and emotional variability of synthesized voices.”

Dubformer leveraged its cutting-edge stack made up of Automatic Speech Recognition (ASR), Text-to-Speech technology (TTS), voice biometrics, machine learning and other tools to deliver this end-to-end solution. A notable aspect of the entire process is the human-curated proof-listening platform, where native-speaking professionals fine-tuned the AI dubbing in the final stages of each project to achieve refined and polished results.

Anton Dvorkovich, the CEO of Dubformer, has commented:

“The broadcast media is actively leveraging YouTube and other UGC (User-Generated Content) platforms, where traditional localization approaches prove inefficient. These platforms require large volumes of content with unpredictable monetization. Our approach empowers media companies to cut video localization costs by 30-50% without compromising on translation volume and broadcast quality.”