Duke Listens!: Visit my main blog at MusicMachinery.com

Stephen Downie gave an overview of MIREX and described the special challenges and issues for this year. He then opened it up to a general discussion. Some comments

• Doug Eck: Make data available, make it easier for machine learning types to participate
• Brian, Daniel M. - can we/should we use more social data in the evaluations
• Some discussion about using virtualization technology to allow the participants to login and run their submissions.
• Geoffry - repeats the suggestion for making subsets of data available. Perhaps people who participate should have to provide some data as well.
• Scheduling - Kris suggests that perhaps we should decouple MIREX from ISMIR.

An interesting discussion - but folks seem to be thinking that MIREX really should be a source of data for researchers - when the role of MIREX is really for evaluations. Anyone who wants to share data should think about setting up a MIRDX (MIR Data eXchange).