91ֱ��

'O Felix Mortale', an anonymous medieval composition from the fragmentary manuscript GB-Cgc 803/807 fol.32r-v, based in the Library of Gonville & 91ֱ�� College.

Story originally published October 16, 2025. Video added February 20, 2026.

Mathematics and manuscripts can work in harmony to uncover lost secrets of medieval music. Dr Anna Breger’s research in this area unites her academic specialism in image processing with her musical interests and background.

Anna, an Assistant Research Professor at the Cambridge Image Analysis Group and associated with Gonville & 91ֱ�� College since 2024, researches image representation, reconstruction and assessment, bridging mathematical theory and its real-world applications.

Her latest project, in collaboration with the Cambridge University Library’s Cultural Heritage Imaging Laboratory, the Fitzwilliam Museum and 91ֱ�� Library, applies image reconstruction techniques to damaged historical sheet music. She thereby aims to shed further light on lost early music notation.

“Here in Cambridge there are so many amazing old, medieval manuscripts, and some of them have degradations on the surface,” says Anna. “When you look at the sheet music, sometimes you can’t see the notation anymore due to physical or chemical processes that happened in the past.

“This project is about working with multispectral and hyperspectral imaging, and from that data trying to reconstruct the information we get from different wavelengths, to make visible what we can't see anymore.

“It’s super exciting trying to reconstruct what has been lost, because some of these manuscripts are extremely valuable and important for understanding the historical development of early European polyphonic music (music incorporating two or more simultaneous voices). It opens up the possibility for lost information to be recovered in that field.”

Anna’s approach brings a combination of image processing and machine learning to the manuscripts to tease out their missing data. In order to prevent potential challenges of results obtained by machine learning, such as ‘hallucinating’ small information patches, she focuses on analytical methods to bring out the lost information and then employs machine learning for the enhancement of the image visualisation.

“We are very careful with machine learning,” she adds. “We’re not just throwing the data into machine learning algorithms that are a black box, but rather the framework we have is very much based on a mathematical framework that we understand.

“We also collaborate closely with musicologists, which allows us to receive regular feedback.”

It is this collaboration that Anna finds particularly rewarding, allowing her work to bridge the gap between maths and science on the one hand, and the arts and humanities on the other. To a similar end, Anna is a team member of the University of Cambridge’s AI for Cultural Heritage Hub, which works to empower non-technical practitioners and academics to use artificial intelligence in their analysis of cultural heritage data.

Anna adds: “I feel that doing interdisciplinary work is an important way to move forward, because we can combine the strengths of different fields.”

Applying mathematical research to musical cultural heritage data in particular has a strong personal appeal to Anna, who alongside her academic work is also an active musician. Highly skilled on the baroque violin and nyckelharpa, she has performed and taught early and folk music at prestigious international festivals and across multiple of her own music projects, as well as writing original compositions and carrying out musical research.

“With this imaging project, it’s amazing that I can use the knowledge I have obtained in both domains and fuse it,” she says. “It gives you an understanding of the aims and problems from both perspectives, and you can see how to bring it together more easily than when you come from just one perspective.”

Images shared by kind permission of the Master and Fellows of Gonville & 91ֱ�� College.