Pangeanic Strengthens the Linguistic Core of the MOSAIC Project with Multimedia Services

With participation in WP2 and WP3,  leading the former and playing a central technical role in the latter, the company has been responsible for building and deploying the linguistic services that enable MOSAIC to operate as a multilingual audiovisual network. 

During the initial phase, Pangeanic established the functional and technical framework that would guide the project's subsequent development. The needs of the European broadcast sector were analyzed, real-world use cases were defined, and technology requirements were aligned with the ecosystem's interoperability, scalability, and sustainability objectives. 

This work laid the foundation for a federated architecture capable of connecting different European nodes while maintaining data sovereignty and technological coherence. 

Strategic Leadership and Definition of the MOSAIC Ecosystem 

During the first months of the project, Pangeanic led WP2 as the strategic axis for establishing the functional, legal, technical, and market foundations of the federated MOSAIC ecosystem. This work made it possible to align technological development with the European regulatory framework and with the real needs of the audiovisual sector, ensuring that the architecture and AI solutions were defined on solid, measurable, and impact-oriented criteria. 

The main lines of action were: 

• Analysis of the European media market and legal-ethical framework: A comprehensive study was conducted on the adoption of Generative AI and Machine Learning across the audiovisual value chain, accompanied by the definition of an intellectual property rights (IPR) management model and a comprehensive ethical framework. As a result, a structured model for the secure management and monetization of content was delivered, along with an ethical framework articulated around eight key dimensions (including bias mitigation, transparency, and data protection), reinforcing content owners' trust in the ecosystem. 
• Definition of the semantic model and organization of multilingual metadata: The EBUCorePlus standard was adopted to structure and harmonize metadata in a multilingual and interoperable environment. As a result, natural language semantic search capabilities and automatic asset retrieval with translation and subtitling in multiple languages were enabled, significantly improving the operational efficiency and accessibility of the archive. 
• Mapping and adaptation of AI tools to real broadcaster needs: Validation and demonstration sessions were held with European broadcasters to align linguistic and AI tools with specific use cases. As a result, 17 improvement requirements were identified and prioritized (classified by priority level), including audio gender detection, identification of multiple languages within a single conversation, and automatic generation of detailed tags to ensure the traceability of the AI methods used. 
• Definition of metrics and impact indicators (KPIs): A technical and commercial evaluation framework was designed to measure the performance and sustainability of the MOSAIC ecosystem. 
  Quantifiable targets were established, including a 30% reduction in content processing costs, validation of more than 15 multimodal workflows, coverage of at least 20 languages and 50 European regions, and a user satisfaction level above 90%. 

Taken together, WP2 not only defined technical and functional requirements, but also shaped a comprehensive strategic foundation for the project's development. Under Pangeanic's leadership, it was ensured that technological innovation advanced in alignment with regulatory compliance, European ethical principles, and commercial viability — establishing a framework of trust and sustainability that facilitates the future scalability of the MOSAIC ecosystem. 

API with Advanced Multimedia Services 

During the prototyping and technical integration phase (WP3), Pangeanic's contribution materialized in a fundamental element: the deployment of a unified linguistic and multimedia services API. This API, offered as a decoupled service securely integrated with the MOSAIC nodes, enables the programmatic and scalable execution of tasks such as:

• Automatic audio and video transcription (Speech-to-Text). 
• Multilingual translation of texts and subtitles. 
• Synthetic voice generation (Text-to-Speech). 
• Automatic summarization and generation of editorial metadata. 
• Automatic subtitling.
• Automatic dubbing processes between languages. 

The approach was service-oriented: each capability is exposed as an endpoint that allows the platform to orchestrate complex workflows combining different tasks without the need to modify the underlying infrastructure. In this way, the MOSAIC ecosystem can be orchestrated in different ways, creating workflows tailored to specific needs. 

Furthermore, the hybrid infrastructure, based on cloud environments with redundant capacity in an owned data center, ensures high availability, elasticity under peak loads, and operational continuity, all of which are essential elements in audiovisual production environments. 

Contribution