When listening to an orchestral performance, we are meant to experience the music as a whole. A well-rehearsed orchestra sounds together, as a unified and synchronised unit (that’s what the conductor is for). By while this is the goal, our auditory system also affords us the ability to tune into certain instruments, or certain voices so that we are able to group many different elements of the orchestra in different ways. For example, when paying attention to what the violin section is doing, we form a group based on timbre similarity. By contrast, if the violins double the clarinets, we could group this section based on melodic similarity. Although different timbres, we recognise that they are moving together in time.
This ability far extends the concert-going experience and is actually a huge part of our auditory system. It even has a fancy name to boot: Auditory Scene Analysis (ASA). This refers to our ability to compartmentalise many sounds into groups, or recognise the single components of an incoming group. ASA has great survival values, and alerts us to dangers such as oncoming traffic, or tigers in the brush ready to pounce.
The term was coined by Al Bregman from McGill University. He has written several books on ASA, and has a nifty website with some nice demonstrations:
The plot below shows the main two ways in which we segregate the auditory scene into different streams.
First, music exists in time, and so we can group musical events as they occur in time. This is known as sequential integration. Consider a rhythmic pattern: our perception of the rhythmic pattern changes according to how the individual parts of the pattern are organised in time. If the time events are too spread out over time (longer than 500 ms between each event), our percept of rhythm is weakened. If the events are too close in time (less than 1000 ms between each event), there is not enough separation between each event to make out the beat properly. Sequential integration also accounts for how we qualify a rhythm’s grooviness. A groovy drumbeat is groovy thanks to minute differences in timing from an ungroovy drumbeat (groovy vs. ungroovy isn’t a very helpful descriptor but I hope you get the idea. I hope one day to write about groove in more depth).
Second, music exists within a very large range of frequencies, and we also have the ability to group sounds according to the spectral relationships, even when they are occurring simultaneously. This is known as spectral integration, and here the focus is on grouping frequencies together to create a sound unit (picture differentiating the between the male and female voices in a choir). Some principles governing the grouping of sound according to frequency spectrum include the following:
- whether the frequencies start and stop together
- whether they are multiple integers of each other
- whether they have the same intensity dynamics (e.g. loudness factor)
- whether they come form the same location in space
(there is a third way to segregate sound, having to do with continuation, but I’ll explain that one in a different post).
We’ll dig into ASA a little further in a future post. For now listen to the example below, listen to the performance of the Bach Partita and count the number of independent lines you hear. You are mostly likely hearing two single voices. Although originating from the same instrument, the voices are segregated because they are occurring far enough in pitch range (sequential integration). If the higher voice sounds more melodic than the bottom voice, it might be because the notes in that voice are spread further apart in time than the bottom voice’s fast-moving figure.