A Framework for Automatic Mixing Using Timbral Similarity Measures and Genetic Optimization