Contemporary Machine Learning for Audio and Music Generation on the Web: Current Challenges and Potential Solutions

Grierson, Mick; Yee-King, Matthew; McCallum, Louis; Kiefer, Chris; and Zbyszynski, Michael. 2019. 'Contemporary Machine Learning for Audio and Music Generation on the Web: Current Challenges and Potential Solutions'. In: ICMC/NYCEMF 2019. New York, United States 16-23 June 2019. [Conference or Workshop Item]
Copy

We evaluate specific Web-based technologies that can be used to implement complex contemporary Machine Learning systems for Computer Music research, in particular for the problem of audio signal generation. As a result of greater investment from large corporations including Google and Facebook in areas such as the development of Web-based, accelerated, cross-platform Machine Learning libraries, alongside greater interest and engagement from the academic community in exploring such approaches, Machine Learning is becoming much more prevalent on the Web. This could have great potential impact for Computer Music research, acting to democratise access to complex, accelerated Machine Learning technologies through increased usability and flexibility, in tandem with clear documentation and examples. However, some problems remain in relation to the creation of more complete Machine Learning pipe-lines for Music and Sound generation. We discuss some key potential challenges in this area, and attempt to evaluate some relevant solutions for developing more accessible Computer Music Machine Learning systems.


picture_as_pdf
ICMC2018-MG-MYK-LM-MZ-CK-CAMERA-READY.pdf
subject
Accepted Version
Available under Creative Commons: Attribution 3.0

View Download

Atom BibTeX OpenURL ContextObject in Span OpenURL ContextObject Dublin Core Dublin Core MPEG-21 DIDL Data Cite XML EndNote HTML Citation METS MODS RIOXX2 XML Reference Manager Refer ASCII Citation
Export

Downloads