Auditory Stream Segregation: Boundaries

Auditory streaming is the ability of the auditory system to group together multiple acoustic events and to assign them to a specific source or “stream.” Music continuously engages these processes of auditory streaming. Not only do the sound frequencies used in music cover the entire human auditory range but music can also be composed with many different simultaneous melodies. The ability to hear those melodic streams is of paramount importance to the enjoyment of music. In a symphony, many musical lines (or streams) can often be heard simultaneously. However, if all the parts were played precisely as notated using a simple synthethizer, as used in the first video games, the pitch of each note would be preserved but their distinctive identities would be lost. This would lead to perceptual fusion of many musical lines, impoverishing the listening experience.

Auditory streaming has been studied in detail by researchers such as Albert Bregman and Leon van Noorden. In a classic paradigm, listeners are presented with a sequence of two alternating tones differing in frequency (ABA-ABA). If the frequency difference is smaller than a first boundary, called just-noticeable difference (JND), the two tones will sound the same. The sequence will be perceived as three repetitions of the same note (AAA-AAA). The JND for frequency ranges between 2 percent and 70 percent of a semitone depending on the frequency and the stimulus duration. Above this boundary, two different tones will be perceived. When the frequency difference is still small, both tones are grouped into a single stream. The sequence is then perceived as a single melodic sequence composed of two notes (ABA-ABA).

Beyond a larger frequency difference, called the fission boundary (FB), the sequence may start to be perceived as two independent sequences of one repeating note (A-A A-A and B---B). This boundary is often estimated around two or three semitones. When the frequency difference near this boundary, the sequence perception is often bistable, as the sequence can be heard as either one or two streams. Just as the Necker cube is an ambiguous line drawing that can be interpreted in two ways, no intermediate percept can be observed. This percept is influenced by central processes such as attention and expectation.

If the frequency separation is larger than a third boundary, called the temporal coherence [Page 106]boundary (TCB), the sequence can only by heard as segregated. However, in a musical context, this boundary is rarely achieved and only occurs at very high tempi. If the tones A and B were eighth notes in a 4/4 meter, the tempo would have to be 500 beats per minute (bpm) for the temporal coherence boundary to reach four semitones. The boundary increases in a lawful manner when tempo is decreased. For example, it reaches six semitones at 300 bpm, and 12 semitones (an octave) at 200 bpm.

Figure 1Three different boundaries reached as a function of the physical difference between auditory streams

Figure 2First three bars of J. S. Bach Partita no.1 in Bb, Gigue

It is reasonable to assume that such boundaries have shaped the conventions and rules of music composition—an idea that was extensively explored by David Huron in 2001. First, the semitone is an interval wide enough to be above the JND. Second, an analysis of the occurrence of the melodic musical intervals reveals that the second major is the most frequently used (about 30 percent), followed by the unison (about 20 percent), the second minor (about 15 percent), the thirds (about 10 percent each), and the fourth (about 7 percent). Larger melodic intervals are rarely used, except for the major fifth. Therefore, the majority of the intervals will fall between the JND and the FB. For the larger melodic intervals, faster tempi need to be avoided to prevent reaching the TCB.

...