Arehart, K., Kates, J. & Anderson, M. (2011) Effects of Noise, Nonlinear Processing and Linear Filtering on Perceived Music Quality, International Journal of Audiology, 50(3), 177-190.
This editorial discusses the clinical implications of an independent research study and does not represent the opinions of the original authors.
The primary goal of most hearing aid selections and fittings is to improve communication by optimizing the perceived quality and recognition of speech sounds. While speech is arguably the most important sound that normal hearing or hearing-impaired listeners encounter on a daily basis, the perception of other sounds should be taken into consideration, including music. Music perception and sound quality is particularly important for hearing aid users who are musicians or music enthusiasts, or those who use music for therapeutic purposes related to stress reduction.
Though some hearing aid users report satisfaction with the performance of their hearing aids for music listening (Kochkin, 2000) the signal processing characteristics that are most appropriate for speech are not ideal for music perception. Speech is produced by variants of one type of “instrument”, whereas music is produced by a range of instruments that create sounds with diverse timing, frequency and intensity characteristics. The perception of speech and music both rely on a broad frequency range, though high frequency sounds carry particular importance for speech perception and lower frequency sounds may be more important for music perception and enjoyment (Colucci, 2013). Furthermore, the dynamic range, temporal and spectral characteristics may vary tremendously from one genre of music or even one piece of music to another. Hearing aid circuitry that is designed, selected and programmed specifically to optimize speech recognition may compromise the perception and enjoyment of music and the effects may vary across musical genres.
A number of studies have examined the effects of non-linear hearing aid processing on music quality judgments. Studies comparing compression limiting and peak clipping typically found that listeners preferred compression over peak clipping (Hawkins & Naidoo 1993; Tan et al., 2004). Whereas some studies found that listeners preferred less compression (Tan et al., 2004; Tan & Moore, 2008; Van Buuren et al., 1999) or longer compression release times (Hansen, 2002), others determined that listeners preferred wide-dynamic-range compression (WDRC) over compression limiting and peak clipping (Davies-Venn et al., 2007).
Arehart and her colleagues examined the effect of a variety of signal processing conditions on music quality ratings for normal-hearing and hearing-impaired individuals. They used simulated hearing aid processing to examine the effects of noise and nonlinear processing, linear filtering and combinations of noise, nonlinear processing and linear filtering. Their study had three primary goals:
1. To determine the effects of these processing conditions in isolation and in combination.
2. To examine the effects of nonlinear processing, noise and linear filtering on three different music genres.
3. To examine how these signal processing conditions affect the music quality ratings of normal-hearing and hearing-impaired individuals.
Subjects included a group of 19 normal-hearing adults with a mean age of 40 years (range 18-64 years) and a group of hearing-impaired adults with a mean age of 63 years (range 50 to 82 years). The normal-hearing subjects had audiometric thresholds of 20dBHL or better from 250 through 8000Hz and the hearing-impaired subjects had sloping, mild to moderately-severe hearing losses.
Participants listened to music samples from three genres: a jazz trio consisting of piano, acoustic bass and drums; a full orchestra including string, wind and brass instruments performing an excerpt from Haydn’s Symphony No. 82; and a “vocalese” sample consisting of a female jazz vocalist singing nonsense syllables without accompaniment from other instruments. All music samples were 7 seconds in duration. Long-term spectra of the music samples showed that they all had less high-frequency energy than the long-term spectrum of speech, with the vocalese and jazz samples having a steeper downward slope to their spectra than the Haydn sample which was mildly sloping through almost 5000Hz.
Music samples were presented in 100 signal processing conditions: 32 separate conditions of noise or nonlinear processing (e.g., speech babble, speech-shaped noise, compression, peak clipping), 32 conditions of linear filtering (e.g., high, low and bandpass filters, various positive and negative spectral tilts) and 36 combination conditions. Additionally, listeners were presented with a reference condition of “clean”, unprocessed music in each genre. Listeners were asked to judge the quality of the music samples on a scale from 1 (bad) to 5 (excellent). They listened to and made judgments on the full stimulus set twice.
The music samples were presented under headphones. Normal-hearing listeners heard stimuli at a level of 72dB SPL, whereas the hearing-impaired listeners heard stimuli as amplified according to the NAL-R linear prescription, to ensure audibility (Byrne & Dillon, 1986). The NAL-R linear prescription was intentionally selected to avoid confounding effects of wide dynamic range compression which could further distort the stimuli and mask the effects of the variables under study.
Both subject groups rated the clean, unprocessed music samples highly. Overall, hearing loss did not significantly affect the music quality ratings and general outcomes were similar between the two subject groups. Average music quality ratings were much higher for the linear processing conditions than for the nonlinear processing conditions. Most noise and nonlinear processing conditions were rated as significantly poorer than the clean samples, whereas many linear conditions were rated as having more similar quality to the clean samples. Compression, 7-bit quantization and spectral subtraction plus speech babble were the only nonlinear conditions that did not differ significantly from clean music samples.
The genre of music was a significant factor in the quality ratings, but the effects were complex, and some processing types affected one music genre more than others. For instance, hearing-impaired listeners judged vocalese samples processed with compression as similar to clean samples, whereas vocalese processed with a negative spectral tilt was judged as having much poorer quality. In contrast, hearing-impaired listeners rated higher mean differences between clean music and compressed samples for Haydn and jazz than they did for vocalese samples, indicating that compression had more of a negative effect on the classical and jazz samples than the vocalese sample.
The outcomes of this study indicate that normal-hearing and hearing-impaired listeners judged the effects of noise, nonlinear and linear processing on the quality of music samples in a similar way and that noise and nonlinear processing had significantly more negative impact on music quality than linear processing did. The effects of the different types of processing on the three music genres was complex and it was clear that different types of music are affected in different ways. Interestingly, these diverse effects were noted even though the music samples in this study were all acoustic samples, with no electronic or amplified instruments included in the samples. The fact that quality judgments of three acoustic genres were affected in different ways by nonlinear signal manipulation implies that the quality of pop, rock and other genres that use amplified and electronic instruments may also be affected in different and unique ways.
Hearing aid manufacturers have begun to offer automatic and manual program options with settings that have been optimized for music listening, though many clinicians may still be faced with the task of customizing programs for their clients who are musicians or music enthusiasts. To complicate matters, the outcomes of this study demonstrate that the optimal signal processing parameters for one genre might not be best for another. In addition, individual preferences could be affected by hearing thresholds and audiometric slopes, though in this study, the hearing-impaired and normal-hearing listeners demonstrated similar preferences and quality judgments, independent of hearing status.
Clearly, more study is needed in this area, but hearing care professionals can safely draw a few general conclusions about appropriate settings for music listening programs. Music spectra contain more low-frequency energy on average than speech spectra, so a flatter or slightly negatively-sloping frequency response with more low and mid-frequency emphasis is probably desirable. As such, music programs may require different compression ratios, compression thresholds and release times than would be prescribed for speech listening. While other special signal processing features like noise reduction, frequency lowering, and fast-acting compression for impulse sounds may need to be reduced or turned off in a music program. These factors combine to suggest a much different prescriptive rationale for music listening than would be require for daily use.
Arehart, K., Kates, J. & Anderson, M. (2011). Effects of Noise, Nonlinear Processing and Linear Filtering on Perceived Music Quality, International Journal of Audiology, 50, 177-190.
Byrne, D. & Dillon, H. (1986). The National Acoustic Laboratories (NAL) new procedure for selecting the gain and frequency response of a hearing aid. Ear and Hearing 7, 257-265.
Colucci, D. (2013). Aided music mapping for musicians: back to basics. The Hearing Journal 66(10), 40.
Davies-Venn, E., Souza, P. & Fabry, D. (2007). Speech and music quality ratings for linear and nonlinear hearing aid circuitry. Journal of the American Academy of Audiology 18, 688-699.
Hansen, M. (2002). Effects of multi-channel compression time constants on subjectively perceived sound quality and speech intelligibility. Ear and Hearing 23, 369-380.
Hawkins, D. & Naidoo, S. (1993). Comparison of sound quality and clarity with asymmetrical peak clipping and output limiting compression. Journal of the American Academy of Audiology 4, 221-8.
Kochkin, S. (2000). MarkeTrak VIII: Customer satisfaction with hearing aids is slowly increasing. Hearing Journal 63, 11-19.
Ricketts, T., Dittberner, A. & Johnson, E. (2008). High-frequency amplification and sound quality in listeners with normal through moderate hearing loss. Journal of Speech, Language and Hearing Research 51, 1328-1340.
Tan, C. & Moore, B. (2008). Perception of nonlinear distortion by hearing impaired people. International Journal of Audiology 47, 246-256.
Van Buuren, R., Festen, J. & Houtgast, T. (1999). Compression and expansion of the temporal envelope: Evaluation of speech intelligibility and sound quality. Journal of the Acoustical Society of America 105, 2903-2913.