Mumbai: NeuralGarage is a technology startup in GenAI & is the brainchild of IIT Kanpur alumni Subhabrata Debnath, Subhashish Saha, Anjan Banerjee, and media & entertainment veteran Mandar Natekar. It is one of the few startups from India that is building core tech with original research and developing its own models in the audio-visual domain. The startup’s flagship technology, VisualDub, addresses a significant challenge in dubbed content — the synchronization of visual and audio cues - ensuring a harmonious viewing experience across languages through visual dubbing and lipsync.
Since its inception in July 2021, NeuralGarage has evolved strategically. After a year dedicated to research and development, and building the technology for commercial use, the company launched VisualDub in May 2023. A successful fundraising round in October 2022 paved the way for building a robust core engineering team, leading to collaborations with more than 10 large global clients.
Building for the world, the company targets a diverse clients & partner base, including film studios, directors, production houses, advertising agencies, OTT players, broadcast networks, and media entertainment companies. Additionally, the company engages with media outlets covering the AI space, technology, and deep tech, establishing a comprehensive outreach and strategy.
NeuralGarage has an ambition to take VisualDub to a global audience by refining its go-to-market strategy and securing further rounds of capital to fuel expansion and innovation. The company has received notable recognition, including being ranked No. 2 at ‘Pitch Fest’ during TechSparks’23 and was a part of YourStory’s Tech30 for 2023. Other accolades include NASSCOM Emerge50 (ranked No. 1 in the media category 2023), Amazon AWS ML Elevate 2023 cohort, and NASSCOM GenerativeAI Foundry. Currently, NeuralGarage has a dedicated team of 20 professionals driving its vision and mission forward.
Indiantelevision.com had the pleasure of conversing with Neural Garage co-founder & CEO Mandar Natekar. In this interview, he offered plenty of valuable insights into the company's operations and vision.
Edited excerpts
On identifying the need for VisualDub in the audio-visual domain
Consider for a moment that you are watching a Korean film with English audio. Your audio input is English while your visual input remains Korean since the visual spoke expression does not match the audio. VisualDub solves for this visual discord, so now when you watch a Korean film with English audio, you feel as if you are actually watching a film shot in English itself. With localization content becoming the biggest currency for the content creation industry, this problem of Visual Discord is amplified exponentially. VisualDub was built to specifically solve this at scale and make content accessible seamlessly to every single person in the world in a language of their choice and which looks truly native and authentic visually in every language.
On the technological innovations behind VisualDub that differentiate it from existing solutions in the market
VisualDub has been built in India by NeuralGarage using its own original research and foundation models and without using third party wrappers. This tech is completely proprietary and is patent filed. It has been built specifically to cater to high impact industries like Films, Streaming, Broadcast Networks and Advertising where preservation of resolution and sync of the original is of paramount important across temporal and spatial movements including background and talent IP. The technology delivers at 2k resolution on pre shot footage without the need to create 3D models or avataars and works across screens of every size from theatrical to mobile phones. This differentiates it completely from any other tech available.
On the role does original research play in the development of NeuralGarage's models for VisualDub
Original research ensures that you are building the tech from a top-down approach, and it’s built with a single focus of specifically meeting the high capability demands of the industries you are building for. Unlike horizontal tech that meets some requirements of many industries with generalized models, VisualDub has been built as vertical SaaS to meet all mandatory requirements of key industries like Films, Streaming Platforms and Advertising. We ensure that we are at the forefront at innovation by staying focused on original research that leads to differentiated tech innovations.
On considering the diverse applications of VisualDub in the entertainment and content creation industry
Currently we are offering it as a service to select large customers to ensure that technology is used correctly to deliver the right results and creates a playbook for the industry. Our ultimate goal is to create a desktop version of our VisualDub that can be licensed across both individual and enterprise level solutions depending on the size of the business and the annual contracts. We are currently raising capital to be able to move in the direction.
On NeuralGarage anticipating VisualDub revolutionizing the content creation landscape and some potential impact you foresee it having on the overall quality of dubbed content across languages
The biggest problem in dubbed content is the visual discord that makes it unrelatable and unauthentic. Content today is monetized across multiple revenue streams, and this is achieved through dubbing. This opens up massive global geographies for content monetization. With VisualDub, dubbed content as we know it ceases to exist. Now with technology it actually gives studios, streaming platforms to create content in every language without incurring additional production cost! VisualDub will allow you to shoot in one language and recreate it authentically in every language on earth without making it look like dubbed content.
On your aspirations in terms of market penetration and becoming the go-to solution for audio-visual synchronization in dubbed content
VisualDub has been made in Bharat for the globe. It solves a very pertinent problem for the entire Entertainment, Media & Content industries across the world and we intend to be the default technology for the post production pipeline across these industries. We are creating a new category “Visual Dubbing” in the post production workflow and VisualDub will define this through original innovation and globally scalable use cases.
