With the introduction of the Walkman® by Sony in the late 1970s, personal music listening habits were forever changed, only to be supercharged early in this century by Apple's iPod® players. The proliferation of personal music players (PMPs) is a cultural change that raises concern because of the tremendous quantity of audio stimulation that users—especially children—can now regularly introduce to their ears.

Much has been reported about the use of PMPs in the scientific and trade literature as well as in the popular media. A Google search for the 12 months from January 2008 to January 2009 for the terms "warning music 'hearing loss'" yielded 19,000 hits. Although a number of these hits represented measured information and advice, many included dire predictions such as this 2009 headline from News.Com.Au: "Experts warn of MP3 player hearing loss 'catastrophe.'"

The purpose of this article is to provide a factual basis to assess the potential risks of PMP use by offering guidance on how audiologists can accurately measure PMP sound levels to estimate realistically the risk of hearing damage. The sidebar and Figure 1 [PDF, 1.1MB] on summarize the potential risks of PMP use relative to other noisy activities for young people. Regardless of whether an audiologist chooses to measure earphone outputs, it is important to understand the relative risk of music listening so that clients may be counseled appropriately.

Although PMPs may not be the bane of young people's hearing, these devices can be played too loudly, too long, and too often. Indeed, studies indicate that most of these devices are capable of producing high sound levels (Fligor & Cox, 2004; Keith et al., 2008; Portnuff & Fligor, 2006). Investigations of typical PMP listening levels suggest that most users adhere to safe levels, although there is evidence that a small percentage do not. Some young people (estimates vary widely) play PMPs at sound levels of 85 dBA or greater especially when background noise is present, using higher volume settings with louder background noise (Airo et al., 1996; Fligor & Ives, 2006; Portnuff et al., 2009). One mitigating factor in the presence of background noise is earphone type. Listeners tend to choose lower output levels in noisy environments when using sound-isolating earphones (Fligor & Ives, 2006).

Only a handful of studies have attempted to factor in PMP use times, which are needed to estimate eight-hour equivalent average exposures. These results suggest that about 15% to perhaps as many as 25% of users would be expected to have exposures equal to or exceeding 85 dBA on an occasional or routine basis (Airo et al., 1996; Portnuff et al., 2009; Williams, 2005). Of concern is that the rate of PMP use is on the rise. Average listening time has increased over the past 20 years, from about 40 minutes per day in the 1980s to an hour per day in the 1990s to two hours per day in this decade (Ahmed et al., 2006; Airo et al., 1996; Bradley et al., 1987; Felchlin et al., 1998; Passchier-Vermeer, 1999; Rice, Rossi, & Olina, 1987; Torre, 2008; Williams, 2005).

Although hearing risk from PMPs may not be a widespread public health concern, further research is underway. It is important for hearing professionals to identify those individuals who are at risk by accurately measuring PMP sound levels to assess more sensibly the likelihood of hearing damage.

Accurate Measurement of Earphone Sound Levels

Measurement of sound levels is easy today because of the wide array of instrumentation available from consumer-oriented suppliers and sophisticated instrument manufacturers. But correct measurement of sound levels—particularly for earphone devices—is another matter. The key to understanding the problem is related to the transfer function of the open ear (TFOE), also referred to as the head-related transfer function (HRTF). Because of resonances of the ear, sound levels at the eardrum will be higher than those measured outside the ear (see Figure 2 [PDF, 1.4MB]).

Sound levels generated by earphones are often measured by placing the earphones on an ear simulation device, such as an artificial head with representative ear canal or perhaps an audiometric earphone coupler. The levels reported are those at the microphone diaphragm in the ear canal, represented by point A in Figure 2 [PDF, 1.4MB]. However, noise hazard, which is based on the relationship between sound levels and noise-induced hearing loss (NIHL), has historically been measured at point B. This disparity is the result of noise-damage criteria derived from studies conducted in the 1960s, when workers in industrialized countries were exposed to noise without the benefit of hearing protection, substantially complicating the matter of estimating effective noise exposure at the ear. At that time, standard practice was to position a sound-level meter in the workplace and record the levels at the worker's location in the absence of the worker. Today's miniature microphones and ear- or shoulder‑mounted instruments, such as dosimeters, were unavailable. Thus risk was, and still is, based on measurements at B, whereas PMP levels are measured at point A.

Because of the ear's natural amplification characteristics, primarily its ¼ wavelength ear canal/tube resonance, the levels in the 2–4 kHz frequency range measured at point A are more than 10 dB greater than those in the undisturbed sound field, measured at point B. Thus, reporting eardrum or coupler sound-pressure levels can substantially overestimate the hazard of sounds, such as music, that contain substantial high‑frequency content.

The preferred method of measuring earphone systems is described in an international standard on the determination of sound immission for sound sources placed close to the ear (ISO 11904, Parts 1 and 2), and a European standard on measuring the sound levels from earphones used with sound system equipment (EN 50332). These methods utilize an occluded ear simulator, a version of which is incorporated into acoustical manikins such as the well-known KEMAR (Knowles Electronic Manikin for Acoustic Research) shown to the left in the photo on pg. 14. The manikin comprises a head and torso with anatomically correct and flexible ear simulations (including the pinna and concha), a cylindrical metal ear canal, and a specially designed acoustic coupler in which the microphone is placed at the termination of the ear canal. Even with this device, care should be taken to correctly position the earphones to reproduce the type of fit found on human ears. Sound-level measurements are critically affected by the earphone fit and seal, especially when the earphone has an earplug-like tip that enters the ear canal (insert) or caps the canal at its entrance (earbud) (Keith et al., 2008). When you fit earphones on your own ears, you can easily hear differences in sound quality and fullness of the bass as you change the earphone fit and seal.

The test signal should also be considered in the measurement of output levels. Signals representative of those listened to by typical PMP users are best for validity of the data, but they can be difficult and time-consuming to measure because of their moment-to-moment variability. Therefore, a stationary test signal such as a shaped pink-noise spectrum is generally employed. The noise is tailored to a frequency spectrum representative of music or speech. One standardized spectrum is the IEC test signal (IEC, 1985) that compares favorably with actual representative music spectra (Keith et al., 2001). Once the measurements are recorded, they are A-weighted (dBA) to properly reflect hearing risk. The second key step is to transform the ear canal measurements to values suitable for noise-hazard analysis, i.e., equivalent sound field. This transformation is done by subtracting the TFOE values, frequency by frequency, from the eardrum-equivalent values that were recorded. Without the TFOE correction, as will be shown below, the measured values would substantially overstate the true noise risk by as much as 15 dB.

Casual Estimates of Sound Levels

Because acoustical manikins are often not available, efforts have been made to estimate sound levels of PMPs in other ways. Such "casual" measurements of earphone levels have frequently been reported in the literature and the media. Measurements have been taken by holding a microphone next to the earphone and inappropriately using those values to represent the risk. Laboratory measurements have been reported that were done incorrectly with inappropriate couplers, or with couplers or probe microphones but without the requisite TFOE corrections. These data appear in some of the earliest publications on the topic of PMPs (Katz et al., 1982) as well as recently in otherwise quite well‑conceived studies (Torre, 2008).

One popular approach to estimating sound levels of PMPs is the Jolene manikin (shown to the right in the photo on p. 14), which is intended to educate youth about safe-listening practices. This inexpensive home-built test device is made from a department store manikin and a Radio Shack sound-level meter. Her construction and use is described in The Jolene Cookbook (Martin & Martin, 2007), available for free download at Dangerous Decibels.

Errors in Casual Measurements

In our experiments, we compared the preferred measurement approach using an acoustic manikin (and transformation of those values using TFOE) to casual measurements with simplified manikins, or no manikin at all and no TFOE correction. We measured the output levels of PMPs with five different insert earphone systems (such as the Etymotic ER6), five earbud-type systems that seal at the entrance of the ear canal (such as the Apple In-Ear Headphones), three supra-aural devices that rest on the pinna (two of which included active-noise reduction and were tested in the on- and off-modes), and four circumaural devices that seal around the pinna against the head (one of which was an active noise reduction device and tested in its two modes).

We took measurements with two different half-inch and two different one-inch microphones held directly against the earphone diaphragms, with small dosimeter microphones in the ears under the devices in two different orientations (for the supra-aural and circumaural earphones) using Jolene; we also took measurements with a few devices (data not shown in Figure 3 [PDF, 1.1MB]) with a standard earphone coupler used for audiometer calibration (NBS-9A). These values, without any TFOE corrections, were compared to what was considered "truth," i.e., data measured using KEMAR with soft pinna and TFOE corrections. For the KEMAR measurements, the microphone output was monitored by the third author while he adjusted the earphones for best fit.

The results are presented in Figure 3 [PDF, 1.1MB], grouped by earphone type. The vertical axis represents the error in the measurement of the overall A-weighted sound level of the IEC-shaped pink noise, compared to the "truth," the TFOE‑corrected KEMAR measurements (i.e., a positive result indicates that the casual method overestimated the actual sound field equivalent level). For the supra-aural and circumaural devices all methods agreed within ± 5 dB, except in one instance in which the one-inch microphone data were high by 10 dB. For the earbuds, the spread was greater with overestimates generally no more than 5 dB, but with one underestimate of almost 15 dB. The insert earphone data were the most variable with errors of + 20 to -15 dB.

When viewing the Jolene data in this chart, note that the instructions for this device suggest a 5‑dB TFOE correction (subtraction from measured level), which was not made for the data in this figure. In fact a 5‑dB correction would closely align Jolene with the "truth" except for the insert earphones. This result was due to Jolene's ear canal, which is short and does not allow for correct insertion of such devices. Thus, she is a fine teaching tool, but her use for determining risk of insert earphones is problematic.

The hearing risks of excessive noise exposure have become increasingly publicized in the past decade. As PMP utilization pervades the culture, warnings about their hazard become more strident. In this report we endeavored to place the risk of PMP exposures in perspective compared to other noisy recreational activities, and to provide guidance on the accurate measurement of the sound levels they create.

It is certainly laudable to educate our youth about the potential risks of PMPs, but exaggeration can diminish our credibility and divert attention from what are arguably more hazardous noisy activities. All efforts to quantify risk of hearing loss from various noise sources, including PMP use, must be based on accurate measurements of both the levels and durations of exposure.

Based on our studies, we recommend the following:

• Use accurate equipment (such as acoustic manikins) and standardized protocols (ISO 11904) for measuring output sound levels of all PMPs (and earphone types) when data are critical for research purposes or litigation.
• Coupler/eardrum/manikin measurements must be TFOE-corrected; otherwise overestimates of 4–8 dB will occur when using resulting sound levels to predict hearing risk.
• Hand-held mics can over- or underestimate levels by as much as 15 dB for insert and earbud devices, but are within about 5 dB for supra-aural and circumaural devices.
• Jolene is a fine teaching tool and her suggested 5-dB TFOE correction is suitable for all types of earphones except insert earphones, for which she is not applicable.