New Life for Old Media: Investigations into Speech Synthesis and Deep Learning-based Colorization for Audiovisual Archives
The Netherlands Institute for Sound and Vision (NISV) is the national audiovisual archive and media museum of the Netherlands. The collections comprise of over one million hours of audiovisual material. One of the collections is that of the so-called “Polygoon” newsreels from the 20th century. This paper outlines recent explorations where Artificial Intelligence technologies are used to enrich this archival material to allow for new types of engagement. Firstly, we investigated leveraging an existing, limited corpus of broadcast narration by a single person to build a working text-to-speech (TTS) system. Secondly, we investigated the possibility of colorization of old black-and-white video footage from the Polygoon newsreel collection using Deep Learning approaches.
|Artificial Intelligence, Audiovisual Archive, Colorization, Deep Learning, Speech Synthesis|
Marsman, R, Karavellas, T, de Boer, V, & Oomen, J. (2017). New Life for Old Media: Investigations into Speech Synthesis and Deep Learning-based Colorization for Audiovisual Archives.