68 Sound Segregation

It is rare that we hear just one sound at a time. Usually, we’re hearing several things—a bird chirping, a car driving by, a conversation on the sidewalk—and we have to separate them from each other in order to make sense of them. This is called  sound segregation. We have several tools to do this.

Schema-based strategies are top-down. A schema is essentially a structure in our brain that holds and organizes the information that we have obtained while growing up. Schema-based strategies are essentially using prior knowledge to locate and understand the distance of the sound. Primitive strategies, on the other hand, are bottom-up strategies: reflexive strategies that help us group sounds together based on similarity or location.

First, there is the primitive auditory stream segregation, where the brain groups the sound perceptually to form a consistent representation of the object from the sound it makes. A good example of this is when we hear an orchestra. As they play, we separate sounds with similar features (ex. the blare of trumpets) from non-like sounds (ex. the whisper of flutes). Grouping sounds by timbre like the trumpet/flute example is one primitive strategy. Other primitive strategies are grouping by pitch or grouping by location.

The other process we can use to identify the information given from a mixture of sounds is the schema-based analysis. This is a top-down strategy in which the brain matches the sensory signal from the knowledge stored in the memory. An example of this is when you hear a noise at the park and recognize it as a birdsong because your schema of what a particular bird sounds like helps you pick out the notes of that one birdsong from all the rest of the noise in the park.

Even with the help of primitive strategies like grouping by similarity and schema-based strategies like recognizing familiar voices, we still have to work to focus on an interesting sound in a complex auditory environment. Imagine talking to someone in a loud environment and trying to hear what they are trying to say. We are able to hear the sound of interest (the other person voice) by focusing intently on it. Picking out a certain sound like in this example is known as the “cocktail party problem.” The cocktail party problem is a hard problem to solve and requires selective attention.