RTVE has awarded Crosspoint the contract to implement RTVE AI archive metadata at scale. The project brings together Crossmedia, Crosspoint’s digital arm, and Amplify technology to speed up documentation, boost discovery, and increase content reuse.
Why this project matters
RTVE’s Documentary Collection keeps growing. Demand is up for archive-based programmes such as Cachitos de Hierro y Cromo, Viaje al Centro de la Tele, and Conexión Vintage. Meanwhile, RTVE Play needs rich metadata to serve viewers better. With this project, the public broadcaster moves closer to its goal of cataloguing 198,220 hours from regional centres.
Scope and volumes
Each year, the service will automatically analyse 28,000 hours of archive content, including:
- 18,000 hours of Spanish audio
- 5,000 hours of image content
- 5,000 hours of multilingual audio
The technology will be fully deployed in September and launched on 1 October. As a result, RTVE will gain faster access to high-quality metadata across its catalogue.
What the service delivers
Audio analysis
- Accurate audio-to-text transcription with punctuation and capitalization
- Named Entity Recognition (NER) for people, places, and organizations
- Keyword extraction aligned with RTVE criteria
- Automatic classification by type (interview, documentary, news, etc.)
- Summaries tailored to RTVE needs
- Speaker gender detection and voice-based identification of public figures
- Semantic segmentation by story and by speaker for news and magazines
Image analysis
- Face recognition of people and public figures
- OCR to extract on-screen text
- Logo and brand detection
- Relevant object detection and scene type identification
- Image descriptions to enrich accessibility and search
- News-program structure detection: section starts/ends, headlines, sports, weather, credits, live signal, correspondent intros, and end-of-show copy
Multilingual analysis
- Transcription and translation of co-official languages (Catalan, Galician, Basque, Valencian, and Balearic varieties) into Spanish—even in bilingual items.
- Proof-of-concept modules for applause/cheering/shouting, violent content and hate-speech detection, visual search, and recognition of musical works.
Technology: Amplify Seiri
At the core is Amplify Seiri, an AI-powered platform that automates and improves audiovisual archive management. Seiri lets teams locate, validate, and reuse content in seconds. It also offers a training portal to adapt models to RTVE’s editorial context and taxonomy.
Tip for readers: Learn more about Seiri on our product page.
Training and support
Crosspoint will train RTVE staff to operate the system independently. A dedicated support team will be available 12 hours per day, with a max response time of 48 hours. This ensures a smooth onboarding and continuous improvement.
Expected impact
- Faster search & retrieval: Rich, consistent metadata improves findability across the archive.
- Higher reuse: Producers and documentalists can repurpose clips in seconds.
- Better accessibility: Transcripts, summaries, captions, and image descriptions support inclusive experiences.
- Operational efficiency: Automated workflows reduce manual tasks while keeping editorial control.
About the partnership
Crosspoint (part of ES Broadcast) leads service coordination and infrastructure. Amplify provides Seiri and GeNews, along with expertise in AI-driven archive automation and content creation. Together, we deliver the human and technological resources RTVE needs to scale RTVE AI archive metadata reliably.
