Music Through Hearing Aids: Perception and Performance

Martin F. McKinney - martin_mckinney@starkey.com

Kelly Fitz - kelly_fitz@starkey.com

Starkey Laboratories

Signal Processing Group

6600 Washington Ave S

Eden Prairie, MN 55433

Popular version of paper 4aPP8

Presented Thursday morning, April 22, 2010

159th ASA Meeting, Baltimore, MD

 

Introduction

 

Music is all around us - we experience it both deliberately and incidentally in many different situations and environments. In normal situations, we process and perceive music effortlessly, even as a "background task". For those with hearing impairment, however, the task of listening to music can prove quite challenging and current hearing aids do not often provide much assistance. Historically, hearing aids have been developed with the primary goal of improving speech perception. Because music signals have different acoustic features than speech signals, the speech-centric processing in today's hearing aids may negatively affect music listening.

 

However, people do not lose their love of music when they lose their hearing, so it is of interest to understand the impact of hearing loss and hearing aid processing on the perception of music.

 

 

Hearing loss

 

Listeners with hearing loss perceive reduced loudness as well as a more rapid than normal growth of loudness with increasing signal level, an effect called recruitment (see Fig. 1). These two effects combine to limit the dynamic range over which hearing-impaired listeners can perceive sound. Because music signals exhibit large variations in level, some hearing-impaired listeners may find quiet portions inaudible and loud portions painful. The predicted loudness of a musical excerpt is shown in Fig. 2 for both a normal-hearing and hearing-impaired listener.

 

 

Figure 1. Abnormal growth of loudness in hearing-impaired listeners. At low signal levels, the perceived loudness for hearing-impaired listeners is diminished. As the signal level grows, perceived loudness grows more quickly for hearing-impaired than for normal-hearing listeners. At very high sound levels, perceived loudness is the same for both hearing-impaired and normal-hearing listeners.

 

 

Figure 2. Predicted loudness for a musical signal for normal-hearing and hearing-impaired listeners. Loudness for the hearing-impaired listener is diminished.

 

 

State of the art in music processing in hearing aids

 

The primary function of hearing aids is to make sound audible without making loud sounds uncomfortable. The process of amplifying quiet sounds more than loud sounds, in order to accommodate a hearing loss, is called dynamic range compression.

 

Because traditional hearing aid compressors were designed to make speech audible, most current aids typically have a "music program" in which they incorporate distinct processing strategies specifically for music listening situations. Different hearing aid manufacturers choose to process music differently because little is known about the impact of hearing loss on music perception. While there is little agreement about the best strategy, most hearing aids apply less compression to music than to speech.

 

Do these different processing strategies restore the richness of music to people with hearing loss? Many hearing aid wearers say "no". They report dullness, loss of presence, disruption of the acoustic space, and difficulty hearing individual instruments. Many people remove their hearing aids when listening to musicothers simply stop listening to music.

 

Figure 3 demonstrates how many hearing aids fail to restore the normal perception of loudness to music. The predicted loudness of a musical passage, as perceived by someone with a hearing loss, is plotted for three different hearing aids. Two of the hearing aids do not restore loudness, and one amplifies the music to a level that is perceived as considerably louder than normal.

 

 

Figure 3. Predicted loudness perception of a musical signal for different hearing aids

 

In addition to overall loudness, it is useful to examine other aspects of the music signal and understand how perception changes as a result of hearing impairment and hearing aid processing. People with hearing loss may pay attention to many of the same features of music as those with normal hearing, but that does not mean they perceive them in the same way. Figure 4 shows how a hearing-impaired listener would perceive the overall brightness (the perception of high-frequency energy) of music after it has been processed by a hearing aid. The plot shows that brightness is severely diminished over some portions of the signal but near normal in other portions, which can make the music sound unnatural.

 

 

Figure 4. Predicted brightness perception of a musical signal for different hearing aids

 

 

Music Feature Analysis

 

In order to quantitatively assess the impact of hearing loss and hearing aid processing on music perception, we need to examine and model the perceptual features of sound that contribute to the overall perception of music. Features of music can be broadly categorized into pitch, timbre, harmony and rhythm classes. We began our current research by examining timbre cues because they are likely to be disrupted by hearing impairment and hearing aid processing.

 

Timbre is the quality that allows listeners to differentiate two sounds with the same pitch and loudness; it is the collection of attributes that allows us to distinguish a violin from a clarinet playing the same note at the same loudness. It is generally thought that our perception of timbre relies both on spectral cues, (the relative strengths of different harmonic frequencies), and temporal cues, (the shape of the note attack and the progression of harmonic strengths throughout the duration of the sound).

 

We have found that people with hearing loss are still able to discriminate musical instrument timbres, though somewhat less effectively than people with normal hearing (Fitz, Burk & McKinney, 2009). A dominant timbral feature for both normal-hearing people and those with hearing loss is sharpness, the perception of relatively strong acoustic energy at high frequencies. Our current research (see Fig. 5) shows that there are differences in how normal-hearing and hearing-impaired listeners perceive sharpness. This is likely due to the fact that hearing loss is often more severe at high frequencies.

 

 

Figure 5. Sharpness rankings for bandpass noise from normal and hearing-impaired participants plotted as a function of frequency

 

 

Music Intelligibility

 

In addition to research on individual features of musical signals, a method or protocol to quantitatively evaluate the reception of music information is needed. It is relatively straightforward to measure speech intelligibility through metrics like the word-recognition rate. Listeners are asked to identify spoken words under various conditions and their recognition rate is taken as a measure of intelligibility. We have no such straightforward method to measure "music intelligibility" in general. Instead, we can break music up into its constituent elements (pitch, harmony, rhythm and timbre) and assess the perception of these elements in a focused and independent manner (McKinney, 2009). Early results indicate that if protocols are tuned well, we can measure deficits in hearing impaired listeners' ability to discriminate pitch and timbre. Other researchers have proposed similar ideas, such as Russo's (2009) functional hearing test for musicians. Collectively, these protocols could be combined to form a general test for music intelligibility. Such a test would guide the development of signal processing strategies for music processing in hearing aids and allow us to make evidence-based decisions on the correct strategies to pursue.

 

 

References

 

Fitz, K.; Burk, M. & McKinney, M. (2009), 'Multidimensional perceptual scaling of musical timbre by hearing-impaired listeners', Proceedings of Meetings on Acoustics 6, 050001.

 

McKinney, M. F. (2009), 'Measurement of music intelligibility in normal and hearing-impaired listeners''Society for Music Perception and Cognition Conference', Indianapolis, IN.

 

Russo, F. (2009), Hearing Loss in Musicians: Prevention and Management, Plural Publishing, San Diego, chapter 8. Towards a Functional Hearing Test for Musicians: The Probe Tone Method, pp. 145--151.