Speech-input for artistic performance

Presentation

Real time speech recognition can fuel artistic creation by providing a medium for arranging a show while it is being performed, and by forcing the artist to improvise when the ASR fails. We propose a software suite designed to integrate speech transcripts and spoken commands into a live performance. We adapt the Kaldi ASR toolkit to show conditions and build a user interface for describing and running a performance. The software is deployed by the Pulso dance company to run its Homeostatsis show.

Reference:

Benoit Favre, Mickael Rouvier, Frédéric Béchet and Rocio Berenguer, Speech Input for Live Performance: An Impromptu Dialogue Between the Computer and the Artist, in Proceedings of International Workshop on Spoken Dialogue Systems (IWSDS) Demo, 2016.

Screenshots

Software

Source code

Source code can be downloaded from the gitlab of LIS.