MUMBAI: An Indian artificial intelligence startup reckons it has cracked one of cinema’s most vexing problems: making dubbed films look authentic. NeuralGarage’s VisualDub technology has been deployed on War 2, releasing on 14 August, to create what the company claims is the world’s first film visually transformed from one language to another whilst maintaining the illusion of native production.
The Bollywood sequel, originally shot in Hindi, has secured a straight film certificate for Telugu distribution—not as a dubbed version but as an ostensibly original Telugu production. Neural Garage co-founder & chief executive Mandar Natekar describes this as a “fundamental shift” in content production and distribution.
The technology addresses a chronic irritant in global film distribution: poorly synchronised dubbing that breaks audience immersion. Traditional dubbing overlays foreign-language audio whilst retaining the original actor’s mouth movements, creating a jarring disconnect that many viewers find off-putting.
VisualDub purports to solve this by digitally altering actors’ facial movements to match the target language’s phonetic patterns, creating the visual impression that performers originally spoke in the dubbed tongue. The result, Natekar claims, allows producers to sell multilingual versions as distinct original films rather than mere translations.
The commercial implications could be substantial. Indian cinema’s linguistic fragmentation has long constrained box office potential, with Hindi films struggling in southern states where Telugu, Tamil and other regional languages dominate. Conversely, southern blockbusters rarely achieve pan-Indian success without extensive dubbing campaigns.
If VisualDub delivers on its promises, producers could command premium pricing for what appears to be multiple “native” productions whilst incurring costs for just one shoot. Natekar suggests this could double or treble pre-release distribution revenues.
The startup, which has garnered backing from Google’s GenAI Accelerator and AWS’s Global GenAI programme, also won TechCrunch’s Battlefield competition in 2024 and this year’s SXSW innovation award. Such endorsements suggest the technology has impressed seasoned investors and technologists.
Yet scepticism is warranted. Previous attempts to digitally manipulate actor performances—from CGI de-aging to deepfake technology—have often fallen into the “uncanny valley”, where near-human animations feel disturbingly artificial. Moreover, the cultural nuances embedded in regional cinema extend far beyond language, encompassing gestures, expressions and performance styles that may prove difficult to algorithmically adjust.
The broader implications stretch beyond Bollywood. Hollywood studios spend millions dubbing blockbusters for international markets, whilst streaming platforms like Netflix invest heavily in local-language content production. A reliable visual dubbing solution could dramatically reduce these costs whilst expanding addressable audiences.
Natekar envisions actors transcending linguistic boundaries entirely: “Hrithik Roshan can now be a Telugu, Tamil, or even a Spanish actor. Tom Cruise in Bhojpuri? Now possible.”
Such grandiose claims invite scrutiny. The proof will lie not in technical demonstrations but in audience acceptance. If War 2 performs strongly in Telugu markets without viewers detecting artificial manipulation, VisualDub may indeed herald a new era in global content distribution.
For now, the technology represents another front in artificial intelligence’s assault on creative industries. Whether it liberates content from linguistic constraints or merely creates more sophisticated fakery remains to be seen. What’s certain is that traditional dubbing studios should be paying attention.