Abstract
The lack of a standard preservation methodology is an important issue that digital speech archives have to face nowadays. Information that needs to be stored requires the implementation of a methodological framework in order to prevent cultural losses. The Centro di Sonologia Computazionale (CSC) of the University of Padova, cooperating with several digital archives (both speech and music) implemented a scientific methodology for the active preservation of audio documents. The framework is still a work in progress and the methodology is continuously improved with new additions, such as neural networks and machine learning-based techniques for detecting the right equalization curves or recognizing discontinuities on the tape. This contribution presents the abovementioned methodology, as well as several informatics tools for the study, the access and the preservation of audio documents.