Feature selection for content-based, time-varying musical emotion regression