Generating a gender neutral cyborg voice for interfaces in games and other audiovisual media, 3 unit self directed course, supervised by Lynn Hughes, Concordia University.
With small datasets of audio recordings I investigated possible methods, machine learning mainly, to generate wave audio files and build bigger datasets to be used in artworks. The datasets were also based on interspecies and gender neutral recordings that were layered and remixed with common editing software. Next steps will be to work with a SampleRNN or RAVE model to do further tests and balance the weight focusing on the voice of ice mainly. After that the audio will be connected to my cyborg music interface/instrument with which I compose music live. The process will be published here later on.
Thanks to Lynn Hughes, Alain Thibault, Jean-François Primeau and CJ Carr/Dadabots, for advice regarding machine learning models and audio software.