Sarvam AI Unveils Real-Time Dubbing System to Preserve Speaker Voice Across Indian Languages

Indian startup Sarvam AI has launched "Sarvam Dub," an AI system designed to transform multilingual video content creation by preserving speakers' original voices during translation.

The technology promises to replace weeks of traditional dubbing work - involving translators, voice artists, and studio time - with a minutes-long automated process that maintains voice identity across language barriers.

Using zero-shot voice cloning and cross-lingual speech models, the system incorporates built-in duration control during speech generation rather than adjusting timing afterward, which the company claims produces more natural-sounding results.

Sarvam has already demonstrated real-world applications by collaborating with IIT Madras to dub technical lectures and helping stream Finance Minister Nirmala Sitharaman's Union Budget speech in Kannada and Hindi - reportedly the first national budget to be *live dubbed using AI*.

The company positions this technology as particularly valuable in India's linguistically diverse landscape, where content frequently needs translation across multiple regional languages and accents.