On February 7, 2023, Italy-based machine dubbing startup, Voiseed, announced it raised EUR 1m (USD 1.1m) in a seed round from LIFTT and the European Innovation Council (EIC) Fund. The investment closed on January 25, 2023.
Voiseed’s co-Founder and CEO, Andrea Ballista, told Slator that Voiseed also previously received EUR 2.1m in grants from EIC Accelerator. He declined to share any financial information relating to the deal as well as growth and revenue metrics from Voiseed.
Ballista said Voiseed plans to use the funds to “further improve its patented AI core technology together with the proprietary emotional multilingual dataset, allowing the company to finalize its voice platform, Revoiceit, our Generative AI Dubbing solution.”
Ballista brings more than 25 years of experience in audio production for the entertainment industry. Having founded game localization company, Binari Sonori, Ballista oversaw the company’s sale to market leader, Keywords Studios, in 2014 and went on to become Keywords’ Global Audio Director.
For his new venture, Ballista has teamed up with Luca Dell’Orto (COO) — described by his co-founder as an entrepreneur and technology enthusiast — and Lorenzo Tarantino (CTO), who has a background in data science for speech from organizations including Swisscom. Voiseed was launched in 2020.
Voiseed’s new investor, LIFTT, is a sector-agnostic VC firm also based in Italy. It has made 32 investments in startups and SMEs since its founding in 2019 and typically invests “multiple times and at multiple stages of the project.” According to its website, LIFTT has invested EUR 0.5m in Voiseed.
Voiseed’s mission is to “revolutionize the dubbing industry,” which Ballista pointed out is “currently characterized by long processing times, high production costs, and significant operational complexity.”
The CEO said LIFTT was attracted to Voiseed’s “unique and disruptive innovation in the Generative AI sector,” adding that “when adopted in the voice sector, [the technology] will allow us to listen to more and more content in more and more languages.”
Ballista also elaborated on the technology at the core of Voiseed: “The patented technology is based on an AI universal voice model and a proprietary emotional dataset, which allow customers to generate all voice types starting from a few seconds prompt and control any expressive style in any language,” he said.
He continued, “This means that the technology is multilingual at core, natively able to speak any language and deal with any phoneme and pronunciation: it currently doesn’t deal with the translation part, that is left to LSPs and Customers.”
Ballista said, to create a voice, the user needs to upload source voices, the source script, and related translations. From there, the platform will produce the target voices in the languages selected automatically. “We are planning further expansions and integrations with APIs and MTs engines,” he said.
Discussing the target market for Voiseed’s platform, Ballista naturally sees big potential in video game customers. Media and entertainment, advertising, marketing, and corporate video localization are other areas where he identifies strong growth potential for Voiseed.