The effect of temporal envelope on melody segregation

Music is often composed of many different interleaved melodies (or streams) played by one or many instruments. In order to appreciate music, listeners must be able to hear each of these streams and separate them from one another. Stream segregation (also known as auditory scene analysis) is mainly based on our ability to group sounds into streams according to their timbre or pitch. This study tests the ability of listeners to segregate a melody based on the temporal envelope of its notes. A psychoacoustics experiment was run on 26 listeners to test whether they could segregate a 4-note repeating melody from an interleaved distracter composed of pseudo-random notes. The spectral and temporal envelope of the melody was constant. The distracter shared the same spectral envelope with the melody, but its temporal envelope was varied throughout the experiment. Results showed that listeners could segregate a melody based only on the temporal cues. However, the dependence of the result on the task and inter-listener variability might help to explain the discrepancies between the results found in studies previously published in the literature.