tranSkriptorium

tranSkriptorium AI is a technology-driven spin-off from the Universitat Politècnica de València, rooted in the PRHLT research center—an international leader in human language technologies and artificial intelligence. Our core innovation is PrIx (Probabilistic Indexing), a proprietary system designed to make complex, non-digitally-native data—such as manuscripts, typewritten texts, and damaged documents—searchable, structured, and computationally exploitable.
Unlike OCR or HTR systems that rely on deterministic outputs, PrIx generates multiple hypotheses with confidence levels, allowing for robust data extraction even in noisy or ambiguous inputs. The system geolocates terms directly in the image, builds semantic relationships between document elements, and enables probabilistic queries that adapt to the variability of historical records.
PrIx supports tasks such as automated transcription assistance, document segmentation, metadata generation, duplicate detection, and the extraction of named entities, patterns, and even musical notation. It integrates seamlessly with existing archival management systems and can be trained with minimal data from the institution’s own collections.
Our technology is currently deployed across Europe and Latin America, serving national archives, academic repositories, and cultural institutions. In some use cases, PrIx has enabled catalog-level structuring of up to 80% of a documentary collection, significantly reducing manual workload while preserving contextual richness.
tranSkriptorium AI operates at the intersection of data science and cultural heritage, with a strong commitment to human-centered, ethical AI that enhances—not replaces—expert decision-making.