Musical scene analysis of normal and hearing-impaired listeners: Instrument identification with concurrent context tones
Despite rapid advances in hearing aid technology, music listening can be problematic for hearing-impaired individuals (Madsen & Moore, 2014). This problem is likely related to poor auditory scene analysis (ASA) abilities. The fundamental importance of ASA in shaping the perceptual experience of music has been acknowledged for a long time (McAdams & Bregman, 1979). It has even been argued that a key component of musical pleasure derives from the successful parsing of musical scenes (Huron, 2001). At the same time, musical scene perception of listeners with moderate to severe hearing loss has not been studied in any depth (although see Kirchberger and Russo, 2015). Therefore, the extent to which hearing impairment affects the parsing of polyphonic musical scenes remains unclear. For instance, can hearing-impaired listeners discriminate instruments or melodies in the midst of an accompaniment?
The goal of this study is to test musical scene perception of hearing-impaired (HI) and normal hearing (NH) listeners. Specifically, this experiment tests the effects of hearing loss and properties of musical masking tones on an instrument and melody discrimination task.
Participants are groups of ten HI listeners (ca. 50 dB HL) and twelve NH listeners (< 10 dB HL).
The experimental stimuli comprise signals (S) that consist of isochronous (250 ms inter-onset interval, IOI) four-note melodies from a diatonic scale. Maskers (M) consist of two triadic tones (500 ms IOI) from the i) piano, ii) cello, or iii) spectrally matched stationary noise. Each trial consists of the mixture (S+M), followed by two alternatives of the signal (S or S’), and listeners decide which one of the two signals was part of the mixture.
In the instrument task, listeners discriminate melodies played by the flute or the trumpet. In the melody task, listeners discriminate between different (order-wise) permutations of sets of tones. For each combination of task and masker, a 2down-1up procedure adapts the signal level to estimate the 71%-correct discrimination threshold, expressed in terms of the signal-to-masker ratio (SMR).
Preliminary results indicate that HI listeners have drastically higher SMRs than NH listeners in all conditions, reflecting worsened musical ASA. NH listeners yield lowest SMRs for the piano masker, suggesting NH listeners’ potential to “listen in the gaps” of the quickly decaying temporal envelope of the piano.
This experiment demonstrates HI listeners’ general difficulty with ASA in musical scenes. Furthermore, it reveals how the properties of masker sounds affect NH listeners’ musical scene perception, deepening our general understanding of musical ASA. The role of the age and musical training of participants is critically discussed.
Huron, D. (2001). Tone and voice: A derivation of the rules of voice-leading from perceptual principles. Music Perception, 19(1):1-64.
Kirchberger, M. J. and Russo, F. A. (2015). Development of the adaptive music perception test. Ear and Hearing, 36(2):217-228.
Madsen, S. and Moore, B. C. (2014). Music and hearing aids. Trends in Hearing, 18:1-29.
McAdams, S. and Bregman, A. S. (1979). Hearing musical streams. Computer Music Journal, 3(4):26-43.