AI dubbing startup UnAI secures Pre-A funding to expand into sports

The focus was on explaining the actual business performance and scalability achieved in the real market rather than the mere possibility of the technology. Emphasis was placed on the fact that the company has secured major customers, including a global OTT platform and the number one player in the FAST market, and that recurring revenue can be generated through these clients. While platform operators spend significant time verifying technology and quality before signing contracts, once selected as a supplier, continuous orders can be expected. Furthermore, a structure was presented to secure revenue based on long-term customer relationships by increasing the scale of transactions through expanding the number of production languages and types of content.

The funds are planned to be invested in developing the proprietary AI engine and expanding global production infrastructure. By increasing development personnel, the company aims to improve the naturalness and emotional expression of synthesized voices and increase productivity by reducing correction work occurring during the broadcast content production process. The production network involving overseas professional personnel will also be expanded. Through the proprietary tool 'Dyna', external personnel will handle translation, voice generation, and inspection, while internal teams focus on process management and final quality assurance. Through this, the company plans to establish a production system that can maintain consistent quality even as supply volume increases.

The starting point was research during doctoral studies on how naturally AI can reproduce human expressions. At that time, while AI technology was advancing rapidly, there were still limitations in reproducing human emotions and subtle expressions. To improve this, an initial model was built based on the Attention model, after which the research scope was expanded to include sound source separation and timbre synthesis.

In this process, it was expected that general narration would be relatively easy to implement while synthesizing the melody and pitch of music, but it was necessary to perform the work of reproducing the complex expressions contained in human speech, including emotions, breathing, rhythm, and context.

The motivation for commercialization was the experience of watching dubbed content with a Canadian wife. Encountering unnatural dialogue and emotional expressions, it was determined that the technology being researched could be applied. Based on this, AI dubbing was chosen as the first business.

In the short term, the company plans to secure additional global OTT and FAST clients and increase the supply volume for existing clients to strengthen the foundation of recurring revenue. To achieve this, they are improving the quality of the dubbing engine and expanding the production system by utilizing professional overseas personnel. They are also pursuing the acquisition of overseas clients for the AI sports broadcasting service 'ASC'. Based on the experience of recording the first commercialization and revenue in Korea in August 2026, the goal is to expand the supply targets to overseas broadcasters and sports content providers and connect them to continuous contracts.

In the long term, the company aims to go beyond dubbing to become a voice AI technology company that naturally understands and expresses human speech. Currently, in the content production process, they generate more than 10,000 minutes of pre-processed voice data every month to use for engine training. Based on the accumulated technology and data, they plan to build a 'Human-like Voice Agent' that responds to Asian languages. The long-term vision is to develop voice AI that communicates naturally with humans by reflecting differences in pronunciation, intonation, and emotional expression for each language.

The audio field has high technical entry barriers, and commercialization is not easy. Securing excellent technology alone does not make a business viable, but if you overcome these barriers based on expertise, you can possess competitiveness that is difficult for other companies to follow easily. For content localization, it is important to have the capability to deeply understand various technologies—such as translation, generation, training/inference, and voice signal processing—and implement them as a single completed service.

If you are preparing to start a business, you must possess industry expertise along with AI utilization capabilities. If you do not accurately understand the problem you are trying to solve, it is difficult to differentiate yourself through effort or ideas alone. I believe we are currently in a period where new markets are being created as these two capabilities combine. However, as companies that have preempted the market accumulate technology and data, it may become difficult for latecomers to enter. There is no absolute time that is too early or too late for starting a business, but there are times when technological changes and market opportunities coincide. If you see a problem that can be solved by combining your expertise with AI, I hope you do not hesitate for too long.