This collaborative project, led by Orange, was dedicated to the numerical edition of documents. After two decades of digitalization of our literary heritage, we now must state that no reliable solution has been yet proposed, in order to pass from the image of the document to a digital version of sufficient editorial quality. OZALID aims at providing a web service that will enable publishers to submit their editorial work to social networks, for correction and enrichment. The contribution of the SITe team concerns the evaluation and the follow-up of the quality of the processed document, from its submission to the social networks to its final version.
Partners: Orange, BnF, Urbilog, Jamespot, i2s, INSA Lyon, Univ. Paris 8, Univ. Lyon 1
Funds: FUI 12
Optical Music Recognition
This work deals with Optical Music Recognition (OMR), in case of monophonic typeset music. The proposed method relies on two separated stages:
- The symbol segmentation and analysis step, designed in order to deal with common printing defects and numerous symbol interconnexions.
- A high-level interpretation step, based on the fuzzy modeling of the extracted information and of musical rules, leading to the decision.
This frameworks allows to deal with symbol variability, the flexibility and the imprecision of music rules, and merging all these heterogeneous pieces of information. Other innovative features are the indication of potential errors, and the possibility of applying learning procedures, in order to gain in robustness. Experiments conducted on a large data base show that the proposed method constitutes an interesting contribution to OMR.
Partner: Télécom ParisTech