Key information:

• The dataset contains audio features and metadata for a million contemporary popular music tracks and is being analyzed by Columbia University's Lab ROSA.
• Currently, services like Pandora rely on trained musicologists to build their music genome. Lab ROSA has been experimenting with machine learning to analyze music tracks.
• The National Science Foundation's Listening Machine Project focuses on analyzing individual sources in sound recordings.
• A mega-sized dataset is needed to reveal problems in algorithms and allow researchers to compare results.
• The Echo Nest's applications use this dataset, and they hope it will strengthen the connection between academic research and commercial development.

Important details:

• The dataset is 300GB in size and contains derived features of a million tunes in database table form, not audio.
• Sample sound can be obtained from 7digital using code provided by Lab ROSA.
• Lab ROSA suggests starting with a subset of around 10,000 songs for a quick taste.
• The need for a large dataset has been a challenge due to the recording industry's stance and expense.