French SME delivering speech recognition and language analysis technology for conversational AI and security content detection applications.
Vocapia Research is a French technology SME specialising in speech and spoken language processing — their name itself signals this focus (from Latin "voca", voice). In their H2020 projects they contributed speech recognition, audio analysis, and natural language understanding capabilities. In KRISTINA they built components enabling a conversational AI agent to understand and interact with humans, while in DANTE they applied language and audio analysis to detect extremist and terrorist content in online media. Their business is turning spoken and written language into actionable data for both user-facing applications and security-critical systems.
Both KRISTINA (human interaction agent) and DANTE (online content detection) required robust speech-to-text and language processing — the common technical thread across their entire H2020 portfolio.
KRISTINA explicitly targeted an information agent with social competence and human interaction capabilities, requiring dialogue management and spoken language understanding.
DANTE focused on detecting and analysing terrorist-related online content and financing activities, applying NLP and audio/text analysis to a high-stakes security domain.
DANTE's scope — online content spanning text, audio, and video — implies cross-modal analysis capabilities beyond pure speech recognition.
Vocapia's two H2020 projects ran almost concurrently (2015–2019), so a clear sequential evolution is not visible — they pursued both threads in parallel rather than pivoting from one to the other. Their earlier engagement (KRISTINA, starting 2015) was in the ICT/human-interaction space, while DANTE (2016) added an explicit security dimension. If a direction can be read from this, it is a broadening from assistive/conversational AI toward security and content surveillance applications — two markets that both depend on the same core speech and language technology stack.
Vocapia appears to be positioning its speech and language technology as dual-use — valuable for both consumer-facing dialogue systems and security-critical content monitoring, a combination that fits well with growing EU investment in both AI assistants and counter-terrorism tools.
Vocapia has participated exclusively as a consortium partner, never as a project coordinator, across both H2020 projects — a clear pattern for a specialist contributor that joins larger efforts to provide a specific technical module rather than to lead strategy. Despite being a small company, they have accumulated 26 distinct partners across 10 countries from just two projects, suggesting they integrate into large multi-partner consortia. This breadth of partners with no apparent repetition points to a company willing to work with new partners and adapt its technology to different project environments.
With 26 unique partners across 10 countries from only two projects, Vocapia has a surprisingly wide network for its size, reflecting the large consortium structures typical of ICT and Security H2020 initiatives. Their network spans multiple European countries but no clear geographic concentration is evident from the available data.
Vocapia occupies a narrow but high-value niche: applied speech and language technology deployable in both civilian AI products and security intelligence systems. Few SMEs can credibly operate across both the conversational AI market and EU counter-terrorism research simultaneously, which gives them cross-domain credibility. For consortium builders needing a speech technology specialist with a security clearance track record, Vocapia is a rare find at SME scale and cost.
