The Build-Up and Resetting of Auditory Stream Segregation: Effects of Timbre, Level, and Abrupt Change

Two experiments explored the effects of timbre, level, and abrupt change on the dynamics of streaming. Listeners continuously monitored 20-s ABA-sequences (1-kHz base frequency; Δf = 4–8 semitones) and reported when the sequence was heard as integrated or segregated. Experiment 1 used pure tones and narrowly spaced (±25 Hz) tone pairs (dyads); both evoke similar excitation patterns but dyads have a “rougher” timbre. Dyad-only sequences induced a strongly segregated percept, with limited scope for further build-up. Abrupt alternations every 13 triplets produced large changes in the extent of segregation. Dyad-to-pure transitions produced substantial resetting, but pure-to-dyad transitions elicited greater segregation than for the corresponding time in dyad-only sequences (overshoot). Experiment 2 examined the effect of varying triplet level. Pure-tone sequences with a maintained 12-dB difference showed similar build-up, but an asymmetry occurred for alternating sequences. Increased-level transitions caused a significant reduction in segregation (resetting) whereas decreased-level transitions had little or no effect. The results suggest that timbre can strongly affect the likelihood of stream segregation, even without peripheral-channeling cues, but moderate level differences do not. Furthermore, abrupt changes in sequence properties have variable effects on the tendency to report segregation—including resetting, overshoot, and asymmetries in the effect of transition direction.