This tutorial was prepared by the American Speech-Language-Hearing Association (ASHA) working Group on Sound Field Calibration of the Committee on Audiologic Evaluation and accepted for publication by the ASHA Executive Board (EB 119-90). The Working Group members are Laura Ann Wilber, chair; William Melnick; Donald E. Morgan; and Patricia G. Stelmachowicz. The Committee on Audiologic Evaluation members are Sandra Gordon-Salant, chair; S. Joseph Barry; Evelyn Cherow, ex officio; Thomas A. Frank; Gregg D. Givens; Michael Gorga; Susan W. Jerger; Sharon A. Lesner; and Robert Margolis. Teris K. Schery, 1988–1990 vice president for clinical affairs, was monitoring vice president. The contributions of Lucille Beck and Jo Williams are acknowledged. This tutorial is not intended to be a standard; rather its purpose is to provide an overview of issues related to sound field measurement and suggestions for clinical practice.
Auditory measurements in a sound field are routine procedures in most clinical audiological settings. For infants, young children and difficult-to-test individuals who will not tolerate earphones, there are few alternatives to sound field measurements for the determination of threshold levels. Sound field measurements also are used to evaluate the real ear characteristics of amplification systems, either with behavioral methods or probe tube measures of insertion gain. Although there are many problems associated with sound field measures, they remain an integral part of many clinical assessments. Unfortunately, at present there is no standard or official guideline for sound field measurement.
Testing in the sound field requires an awareness of the limitations imposed by the characteristics of the room, the background noise levels, properties of the loudspeaker, unavoidable movement of the listener, the type of stimuli, and a variety of other factors. In some instances it may not be possible to alleviate or even minimize the effects of these variables. The above variables in the acoustic environment may interact unpredictably to affect the reliability of both threshold and supra-threshold sound field measures. From a practical standpoint, clinicians often may be forced to set priorities and compromise accuracy in order to achieve a particular goal. The specific application (e.g., sound field threshold estimates in infants, supra-threshold hearing aid measures) may dictate the decision.
The purpose of this tutorial is to provide an overview of the problems encountered with auditory measurements conducted in the sound field and to furnish suggestions that will significantly reduce or eliminate these problems. Many of the issues addressed in this document are also applicable to probe tube microphone measures of hearing aid gain and output. However, the additional complexities related to stimulus selection and calibration of probe tube measures prohibit a detailed treatment of probe tube measurement within the scope of this paper. The issues that are addressed are related to equipment, sound field environment, selection of test stimuli, calibration of the sound field, and equivalent threshold sound pressure levels (ETSPLs). Finally, this tutorial includes a glossary of terms found in the text, and which often are used in acoustic measurements conducted in the sound field.
The equipment used in sound field measurements consists of a stimulus generator, loudspeakers, and calibration equipment. The stimulus generator is usually an audiometer and will not be discussed here. The problems related to loudspeakers and sound field measurement procedures will be discussed below.
Testing in the sound field requires use of a loudspeaker to transduce an electrical signal to an acoustical stimulus. Problems not typically encountered with earphones may be introduced when using loudspeakers, primarily due to the interactions among the transducer (loudspeaker) characteristics, the acoustic environment (test room), and the positioning of the listener in the sound field. The interactions among these variables affect the spectral characteristics of the acoustic energy reaching the ear of the listener. There exist no American National Standards Institute (ANSI) or International Electrotechnical Commission (IEC) standards defining minimum characteristics to be met by loudspeakers used in audiologic testing; therefore, it is important to review some of the factors affecting loudspeaker performance.
The ideal loudspeaker for audiologic testing should possess the following general characteristics: (a) broad bandwidth (minimally 100–10,000 Hz); (b) constant output as a function of frequency (“smooth” frequency response); (c) low distortion; (d) capability of accurately transducing transient as well as steady-state signals; (e) uniform radiation pattern in the sound field; and (f) high electroacoustic efficiency (high acoustic output with low voltage input).
No single loudspeaker has been designed that can satisfy all of the above requirements. Of the many loudspeakers available, the most widely used is the direct radiator type. Within the frequency and intensity requirements for audiologic testing in the sound field, such loudspeaker systems provide the most efficient transduction of currently employed acoustic signals. However, to attain the desired bandwidth and constancy of frequency response at test locations in the sound field, it is typically necessary to utilize two or more loudspeakers coupled with a crossover network, dividing the overall frequency range so that each loudspeaker operates within its optimum frequency band. Unfortunately, such loudspeaker arrays are particularly vulnerable to sharp changes in output at the frequencies coinciding with the crossover regions. Loudspeaker systems used in sound field testing must be evaluated in the environment in which they will be used to determine the effects of crossover networks on the frequency response of the system.
The frequency response of a loudspeaker may vary greatly in both the overall frequency range and the relative levels within the overall range. Figure 1 includes the frequency response of a typical supra-aural earphone and two hypothetical frequency responses of loudspeaker systems having a nominal frequency range of 200 to 10,000 Hz. Transduction of very narrow band signals probably would not be affected by the variations in frequency response shown for the loudspeakers. However, when complex signals are transduced, the overall sound pressure level (SPL) and the relative contribution of any frequency band in the signal will be affected by the specific frequency response differences across the two systems. For example, if a broad band signal such as speech is transduced through the two loudspeakers the overall SPL might be identical from each speaker, but the relative contribution of low-frequency and high-frequency components could be significantly different between the two loudspeakers. The difference in relative contribution between the high- and low-frequency bands could influence hearing test results when individuals with sharply sloping hearing loss configurations are evaluated. It has been suggested that loudspeakers used for the transduction of speech vary no more than ±5 dB across the frequency range from 200 to 6000 Hz (Dirks, Stream, & Wilson, 1972).
The loudspeakers in Figure 1 also provide examples of irregularities in frequency response which would affect narrow-band signals (but not pure tones) transduced into the sound field. When a frequency modulated (FM) signal (see section on Stimuli) is transduced, the deviations in the frequency response of the loudspeaker interact with the FM signal deviations to introduce amplitude modulation into the acoustic signal. Such amplitude modulations can be avoided by ascertaining that the loudspeaker frequency response is constant within the frequency modulation range of any FM signal to be delivered to the loud speaker. When frequency-by-frequency differences in loudspeaker responses are encountered, it may be possible to adjust the level in the sound field using narrow band equalizers to overcome inadequacies in the loudspeaker system.
When a signal is transduced through a loudspeaker, a near field and a far field are created. Fundamentally, the size of the loudspeaker diaphragm determines the distance at which a theoretical near field/far field boundary will occur. In the near field of any loudspeaker, large SPL changes can occur for small changes in distance from the loudspeaker. It is for this reason that routine measurements are made with subjects located between the near/far field boundary and the direct/reverberant field boundary (see Environment section). However, for special circumstances (e.g., some probe microphone measurements) it may be necessary to place the subject in the near field of the loudspeaker. When a listener is placed in the near field care must be taken to ensure constancy of head position to decrease the influence of acoustic turbulence present in the near field of the loudspeaker. When the far field boundary has been reached the inverse square law applies (i.e., for every doubling of distance a 6 dB decrease in SPL occurs). The implication of the near field/far field principle is that there is a distance from the loudspeaker within which head position (and movement) must be precisely controlled in order to predict accurately the SPL delivered to the ear. In principle, the smaller the diameter of the loudspeaker, the closer to the source at which the near field/far field boundary occurs for any given frequency. Therefore, the near field/far field boundary will, indirectly, determine the size of a loudspeaker required for a test room of any given dimensions. However, when the test position in the sound field is beyond the near field/far field boundary, and the test room is less than anechoic, then the SPL will be affected by whether the position is in the direct or reverberant field (see Environment section).
Testing in the sound field introduces several variables not encountered under earphone test conditions. Because all signals transduced into the sound field will be more complex than pure tones, the effects (as described above) on the test signals of the frequency response of a loudspeaker system must be evaluated. Moreover, the amplitude of the signal at the ear of the listener will be affected by changes in distance between the listener and the loudspeaker. Additionally, the interactions between the unique characteristics of the loudspeaker and the test environment in which the procedures will be conducted must be evaluated.
The purpose of this section is to describe the equipment needed to accomplish measurement of test stimuli and ambient noise within the test environment. For precise measurement of SPL, a sound level meter (SLM) or a measuring amplifier is required. Each of these units is a device which, when used with the appropriate microphone, allows the direct measurement of acoustic signals in SPL.
Measurements in a sound field usually are made with an SLM in conjunction with an octave-band or a 1/3 octave-band analyzer. The SLM used should meet or exceed the specifications listed as Type 1 Sound Level Meters (ANSI S1.4-1983). The octave or 1/3 octave-band sound analyzers should meet the specifications outlined for these devices by the American National Standards Institute (ANSI, S1.11-1986).
The measurement of background noise in a test room may be affected occasionally by the electrical noise of the SLM itself. When this is suspected, the internal noise of the SLM should be evaluated using an isolator cavity or a dummy microphone. If the internal noise of the SLM is not at least 6 dB less than the desired measurement levels, a correction for internal noise will have to be made. The procedures and corrections for internal noise usually can be found in the user's manual for the SLM or can be obtained from the manufacturer. If the measured internal SLM noise exceeds the minimum permissible ambient noise levels (Appendix B, ANSI S3.1-1977), then that SLM should not be used for measuring ambient noise levels (ANSI S12.6-1984).
A condenser or an electret microphone, usually 1/2 or 1 inch, is used to transduce the acoustic signal for measurement in SPL. When the microphone is of the free field type, SPL is read directly from the SLM. If a pressure microphone is used, then appropriate corrections must be made. These corrections, as well as information regarding angle of incidence and compensations for diffuse sound field, are usually available from the manufacturer of the SLM.
Prior to making measurements, a pistonphone or sound level calibrator is used to calibrate the entire measurement system including the microphone, SLM, amplifier or voltmeter. The pistonphone can be used to check the calibration of the SLM only at low frequencies, whereas the sound level calibrator may be used to perform the same function at one or more specific frequencies.
In addition to measures of overall SPL, it is essential to measure the spectral characteristics of the test stimuli. To accomplish this analysis acoustically, a real-time spectrum analyzer or a wave analyzer is used with an SLM to measure the signal. The precision with which the spectrum can be defined will depend on the bandwidth resolution of the measuring equipment. In general, the bandwidth of the measuring equipment should be narrow relative to the nominal bandwidth of the signal to be analyzed.
Sound field measurements are influenced by the acoustic characteristics of the environment in which auditory measures are to be conducted. Ideally, an anechoic chamber would be used for all threshold measurements. However, this is clearly not practical, therefore, it is important to identify the factors that influence the accuracy of measures in the sound field. A number of variables can influence the sound as it is transmitted from its source to the ear of the listener. The spatial pattern of sound around a source in an enclosed space, such as a room, differs from that in an unobstructed sound field. Thus, the inverse square law (decrease of 6 dB for doubling the distance) may not predict precisely the sound pressure as a function of distance from the sound source because of the influence of boundaries and/or obstructions. Therefore, rather than relying on the prediction of the inverse square law, the SPL must be verified at the test position in the room. The volume of air in the room is an acoustic element with natural modes of resonance similar to that observed in structures such as organ pipes. The resonant frequencies of a room are determined approximately by its dimensions. These resonant properties will enhance sound pressure at some frequencies but not at others. The closer the frequency of the exciting sound source is to a resonant frequency, the greater will be the influence of resonant vibration. When pure tones are introduced into the sound field, the resonances of the room are evidenced by standing-wave patterns with resultant variation in measured SPL depending on measurement location in the room.
The number of resonant modes of a room is theoretically infinite. For complex stimuli, amplitude variation due to room resonance is most noticeable at low frequencies. As the frequency of the sound source increases, the resonant peaks move closer together and the resonant effects tend to overlap, resulting in relatively uniform frequency transmission characteristics (Knudsen & Harris, 1978).
The resonant properties of a room result mainly from the reflection of sound at the boundaries of the enclosure. The SPL at any given point in the room enclosure will depend on the source intensity, the directional characteristics of the source, and the acoustic absorption of the room's boundaries as well as the objects within the room. Sound reflections give rise to another acoustic phenomenon which can influence auditory measurements in the sound field: reverberation. By altering the signal-to-noise ratio in the sound field, reverberation can affect measures of absolute threshold as well as measures of speech intelligibility.
A room with poor absorption characteristics is said to be reverberant. This type of room may have large surfaces composed of reflective materials such as glass windows or doors, or may contain reflective objects such as metal cabinets, chairs, and tables. In a reverberant sound field, the reflected waves occur in such rapid succession that the listener does not hear the waves as distinct repetitions of the original sound (echoes), but as a prolongation of the original signal following its cessation. This “prolongation” is expressed in terms of reverberation time. Reverberation time is “the length of time in seconds it takes for the energy in the steady-state sound field in a room to decay by 60 dB after the source of sound excitation is suddenly turned off” (Beranek, 1988, pp. 781–881). Reverberation time is dependent on the geometry of the room as well as the physical properties of its boundaries. In most sound-treated test rooms, the reverberation time is short (0.1–0.2 sec) and is of little consequence. Nevertheless, the reverberation characteristics of the test room are important variables for sound field testing and should be available when the test site is chosen (Hirschorn & Singer, 1989).
The reverberation properties of the test room also dictate the relation of the direct sound field to the reverberant sound field. The direct sound field is dominated by the incident sound wave arising from the source. As such, the SPL as a function of distance from the source approximates that predicted by the inverse square law, and the spectrum is relatively unaffected by reflected sound. The reverberant field, on the other hand, is dominated by the reflected sound waves. Distance from the source becomes less of a factor in determining the existing sound pressure as the sound field becomes more diffuse and less directional. The direct-reverberant field relationship becomes an important consideration in selecting the placement of the listener in the sound field.
The spectrum of the stimulus in a sound field can be affected by reflecting surfaces in the room. Chairs, tables, mirrors, cabinets, and audiometric equipment all represent potential reflecting surfaces. Whether an obstacle represents a significant reflector depends on the composition of its material and its size. If an object is small relative to the wavelength of the propagated incident sound (1/4 to 1/5 of the wavelength), it is usually of little consequence.
In addition to the physical properties, the arrangement of the objects within a room in reference to the sound source can be a significant variable. Changing the arrangement of room furnishings and other equipment in the test room should not be done arbitrarily. The number of participants present during testing should not differ appreciably from the number of participants present when the sound field is calibrated. The presence of an additional assistant or parents or any other observers within the sound field enclosure may lead to erroneous sound field measurements (Dirks et al., 1976). The procedure for measuring the characteristics of the test room is described below.
Testing in the sound field typically involves the presentation of some form of frequency specific stimulus and/or speech materials. Each of these types of stimuli presents a unique set of problems associated with delivery in the sound field. As suggested previously, the problems associated with the delivery of frequency specific stimuli are related to the interactions between room acoustics and loudspeaker characteristics. In fact, because of the unpredictable effect of listener movement, head/body baffle and diffraction, and room acoustics, it is not possible to deliver pure tone stimuli in most sound fields and specify the SPL at the ear of the listener. As a consequence, it is necessary to employ frequency specific stimuli somewhat more complex than pure tones in order to maintain a uniform sound field at the ear of the listener.
The problems associated with the delivery of speech in the sound field are related to specifications of the SPL of transient signals, and typically are addressed in terms of the peak or long-term root mean square (rms) level of the stimulus. The difficulty in specifying speech level is not unique to sound field delivery. The problems are identical whether the speech is to be transduced by an earphone or a loudspeaker. In this section, the problems associated with delivery of frequency specific stimuli and speech stimuli in the sound field are discussed.
The choice of frequency specific stimuli to be used in the sound field requires an understanding of the performance of such stimuli in typical audiometric test environments. Dillon and Walker (1982) have provided a concise summary of desirable properties of frequency specific stimuli. These are listed below:
The stimulus must be frequency specific within the physical limitations imposed by room acoustics and the psychoacoustic limitations imposed by the encoding characteristics of the human ear. Pure tone stimuli best meet these criteria when the environment is anechoic, but are compromised by the effects of standing wave patterns in the sound field (see Dillon & Walker, 1982a for a review; Morgan et al., 1979). Therefore, it is necessary to use stimuli with broader bandwidths than pure tones. It is important that the bandwidth of the signal be limited to a frequency range within which thresholds of normal hearing subjects will not differ significantly from those obtained for tonal stimuli.
The SPLs generated by the stimuli in the sound field must be stable in the region within which the head can move during testing. Dillon and Walker (1982b) estimate that even with the trunk immobile, the adult head can move approximately 20 cm, front-to-back and 8 to 10 cm, side-to-side. With young children, it is virtually impossible to guarantee head stability within a 20 cm range. Therefore, the stimulus of choice will represent a compromise between a signal that provides satisfactory sound field stability with acceptable frequency specificity.
The SPLs generated in the field should be stable for small shifts in frequency. The signal generator itself may “drift” up to ±5% in frequency. The combined interactions of the test room, the loudspeaker, the signal generator drift, and the listener placement in the field may affect the SPL at the test position. Therefore, the signal should not be so frequency specific that such variations in frequency affect calibration or threshold measurement accuracy.
Finally, the stimuli used for measurements in the sound field should enable measurements comparable to those obtained with pure tones under earphones. This will allow hearing loss to be defined accurately within the limits imposed by the two test modes.
The most commonly proposed alternatives to pure tones include frequency modulated (warble) tones, narrow bands of noise, and amplitude modulated tones (Dillon & Walker, 1980; Dillon & Walker, 1982a, b; Goldberg, 1979; Lippmann & Adams, 1982; Walker & Dillon, 1983; Walker et al., 1984). Each type of signal has potential advantages and disadvantages for sound field threshold measures. The report of Dillon and Walker (1982b) includes a thorough analysis of most of these signals. Figures 2, 3, and 4 provide both spectral and temporal representations of the characteristics of each of the signals to be discussed in the following sections.
1. Frequency modulated (FM) tones. Figure 2 illustrates an FM signal. There are three characteristics by which FM tones are defined: (a) carrier frequency, (b) modulation rate, and (c) frequency deviation. The carrier (center or nominal) frequency may be modulated by a variety of waveforms, including a square wave, a triangular wave, a sine wave, or a ramp. Staab and Rintelmann (1972) reported that many (approximately 50%) audiometer manufacturers at that time employed rectangular modulation waveforms, although there existed no psychoacoustic data to suggest an advantage of one waveform over any other. However, Dillon and Walker (1982b) and Walker (1988) suggested that sinusoidal or triangular modulation would probably be best, with a slight advantage to triangular modulation. Second, modulation rate, which is expressed in Hz, refers to the number of tlmes/sec the frequency is swept. The modulation rate is the frequency of the modulation waveform. Third, the frequency deviation defines the frequency range over which the instantaneous frequency varies during each modulation. The frequency deviation may be expressed either as percent of frequency change, from the nominal frequency, or as the frequency deviation in Hz. The bandwidth of an FM signal is dependent on these characteristics. If percent bandwidth is held constant the absolute bandwidth will increase as a function of the nominal frequency; conversely, the percent bandwidth will decrease with an increase in the nominal frequency when the absolute frequency deviation is held constant. Generally speaking, the bandwidth of the FM signal will be dependent on the frequency deviation, as long as the modulation rate is less than the overall frequency deviation. For a more complete description of FM modulation see Malvino (1973). For a discussion of conditions under which the modulation rate and frequency deviation interact to affect the bandwidth of the signal, see the review by Dillon and Walker (1982b) and by Walker (1988).
2. Narrow band noise (NBN). The NBN signal (see Figure 3) is generated typically by passing the output of a white noise or broadband noise generator through a band-pass filter, and then to a transducer. Generally, NBN stimuli have been limited by problems associated with the rejection rate of the filters used to define the bandwidth of the NBN. That limitation will be overcome when the industry routinely employs currently available digital technology in clinical test equipment. This is already done in some computer-based audiometers.
3. Amplitude modulated (AM) tones. An AM signal (see Figure 4) occurs when the amplitude of a high frequency stimulus (carrier) is controlled by a low-frequency stimulus (modulator). A simple (sinusoidally) amplitude modulated tone is represented spectrally as a three-tone complex. The side bands of the center frequency are dependent on the modulation frequency. If the input spectrum to the modulator consists of components fL and fH, the output spectrum will have components at fL, fH, fH-fL, and fH+fL (see Figure 4).
Interactions among signal, test room, and psychoacoustic effects must be considered. As stated above, Dillon and Walker (1982b) described the advantages and disadvantages or limitations of a variety of stimuli for use in the sound field. Many of their findings are summarized in the following sections.
1. Physical effects. When a sound field is not anechoic, as is the case with virtually all clinical test rooms, the introduction of pure tones in the field results in the production of standing waves. As such, the sound field will have areas of minimum (node or null) and maximum (antinode or loop) SPLs throughout the room.
Further, these areas or positions will vary as a function of frequency. The best way to overcome the effects of standing waves is by using a complex signal the bandwidth of which exceeds the bandwidths of the nodes and antinodes. Effectively, the wideband signal enables an “average” measure of SPL, thus minimizing the effect of standing waves on overall SPL. Therefore, for any measurement point in the sound field, signal level variability decreases as the stimulus bandwidth increases. That is, a more uniform sound field is produced by wider bandwidth stimuli.
Additionally, certain frequency specific stimuli produce a more uniform field than others. FM tones result in a more consistent sound level throughout a room than do AM tones of comparable, narrow bandwidth. The inadequacy of the AM tones is predictable because the spectrum of a sinusoidally AM tone consists of only three components, thus limiting the effective averaging over a range of frequencies in the measurement process. The uniformity of the NBN signal is limited in low frequencies due to the amplitude fluctuations of individual components over time. These physical effects notwithstanding, the report of Dillon and Walker (1982a) reveals that when bandwidths are comparable, uniformity of the sound field is constant, at least for FM tones and NBN. The early report of Morgan and coworkers (1979), suggesting a physical superiority in field uniformity of NBN over FM tones, was based on comparison of a 10% bandwidth FM tone with a 23% bandwidth noise. Dillon and Walker (1982b) have clarified the importance of bandwidth and room acoustics on these types of signals.
2. Psychoacoustic effects. The perceptual capabilities of the human listener will affect decisions regarding the signal of choice for sound field testing. For narrow bandwidth signals at low frequencies, there will be an increased variability in threshold estimates (or loudness judgments) due to the short averaging time of the human ear (<200 msec) and the relatively long presentation time (approximately .5–1.5 sec) of a signal whose intensity is varying. Variability is greatest for NBN signals of low frequency and/for FM tones of narrow bandwidths.
Hearing loss also will impose unique limitations for certain frequency specific stimuli. Many persons with sensorineural hearing loss evidence rapid changes in sensitivity with small changes in frequency (e.g., sharply sloping, high frequency hearing loss). As stimulus bandwidth increases and/or filter rejection rate decreases, the correlation between the threshold estimates for that stimulus in the sound field and threshold estimates made with pure tones under earphones will decrease. The hearing loss at a given frequency may be underestimated substantially in cases where threshold sensitivity changes rapidly as a function of frequency (Orchik & Mosher, 1975; Stephens & Rintelmann, 1978). For example, in the case of a steeply sloping hearing loss the listener may respond to frequency components that are remote from the nominal frequency. The magnitude of the error at any given frequency will depend on the signal bandwidth, the rejection rate of the filter used to limit the bandwidth, and the slope of the hearing loss. NBN signals are particularly susceptible to this limitation, because the slope of the skirts of these signals is typically more gradual than for FM or AM tones.
Additionally, short duration signals may influence threshold estimates for normally hearing and hearing-impaired listeners differently because temporal integration can be affected by sensorineural hearing loss. Historically, hearing loss has been assessed using signals of sufficient duration to exceed the temporal integration limitations of the human ear. The use of sound field stimuli of significantly shorter duration than those employed for earphone measurements would introduce unnecessary variables into the test protocol.
3. Stimuli of choice—FM tones. The problems inherent in choosing a particular frequency specific signal for use in the sound field are summarized by Walker, Dillon and Byrne (1984), who reviewed the dilemma imposed by: (a) the need for increased bandwidth in the signal to ensure uniformity of the sound field; and (b) the need for frequency specificity in the signal to ensure accuracy of audiogram slope measurement. Their conclusion is that measurement of threshold in the sound field among persons with hearing loss may require a set of signal parameters chosen to cover a range of hearing loss configurations and/or test conditions. Under most clinical settings it can be concluded that: (a) AM tones are inadequate due to the absence of a uniform sound field; and (b) NBN signals are limited in use by amplitude fluctuation in the low frequencies and rejection rate limitations in the high frequencies. However, the rejection rate limitations of NBN can be solved by use of digital filtering and synthesis techniques.
When the bandwidth of NBN or FM tones is identical, comparable uniformity of sound field and threshold measurement efficiency can be accomplished. However, considering the limitations imposed by most clinical equipment available today, conventional audiometry can best be approximated under sound field conditions by using FM tones having the specific stimulus parameters described below.
4. Modulation waveform. The modulation waveform defines the manner in which the instantaneous frequency of the signal changes relative to the nominal frequency. The waveform may be modulated using sinusoidal, triangular, ramp or rectangular modulation. The disadvantage of rectangular and ramp modulation lies in the considerable energy “splattered” to frequencies beyond the nominal signal frequency (Watson & Gengel, 1969; Wright, 1968). Either a sinusoidal or triangular waveform appears to provide an adequate FM tone for audiometric testing.
5. Modulation rate. The modulation rate or frequency (specified in Hz) refers to the rate at which the frequency range is swept. Very rapid modulation rates result in decreased field uniformity and very slow modulation rates interact with temporal integration properties of the ear to affect threshold sensitivity. Walker and Dillon (1983) concluded that the commonly used 5 Hz modulation rate available in many clinical audiometers may be adequate for normally hearing individuals, but that persons with hearing loss and concomitant changes in temporal integration require a rate of approximately 20 Hz. They caution that increasing the modulation rate unnecessarily will reduce the slope of the skirts of the FM tones and increase the sound pressure variability in dB of the stimulus in the field. They suggest 20 Hz as an adequate compromise.
6. Bandwidth. The most important parameter to consider in the choice of the FM tone is overall bandwidth. This characteristic defines the overall frequency deviation from the nominal (center) frequency. As specified earlier, a more uniform sound field will be obtained with wider bandwidths, but greater error will occur in threshold measurements for individuals with sloping, high-frequency hearing losses. Walker, Dillon, and Byrne (1984) suggest a range of bandwidths for FM stimuli to be used in sound field audiometry. These are shown in Table 1.
Walker, Dillon, and Byrne (1984) conclude that under conditions in which sound field uniformity is the critical criterion (as with a constantly moving infant or young child), “wide” bandwidth stimuli should be used. When accuracy of audiometric configuration is the critical criterion to be met, they suggest that “narrow” bandwidths be used.
Walker, Dillon, and Byrne (1984) recommend that neither the percent deviation from center frequency nor the absolute bandwidth in Hz should be constant for the range of center frequencies required in sound field audiometry. There seems to be no justification for a constant percent deviation or a constant bandwidth for all frequencies if the goals of uniformity of sound field and accuracy of audiometric testing are both to be met. It is probably unrealistic to suggest that several bandwidths of stimuli at each frequency be available in clinical instruments. Accurate identification of rapid changes in hearing sensitivity as a function of frequency is a high priority in any clinical audiologic setting. Thus, when only one bandwidth at each frequency can be included, it is suggested that the “narrow” bandwidths proposed by Walker et al. (1984) be used.
For most sound field measurements the ideal test environment would be provided by an anechoic chamber. These chambers are expensive and usually not available to audiologists providing clinical service. Generally, clinical facilities have used “sound treated” rooms for sound field studies. Sound-treated rooms usually are designed to attenuate environmental sound and provide a quiet environment within which to conduct measures of audition under earphones. These sound rooms are not optimal for sound field use. Before sound field measurements are conducted, the acoustic characteristics of test stimuli must be determined within the room. In addition, the effects of the reflective properties of the ceiling, floors, and walls of the room on the acoustic signal must be defined and ambient noise levels must be measured. Without this type of assessment and calibration of the sound field the measures obtained are of questionable value (Nabelek & Nabelek, 1985).
The measurement of SPL at the point in the sound field where the observer's head is to be placed typically is referred to as the substitution method. Specifying SPL in this way, however, will not reflect accurately the levels arriving at the ear of the observer. Three major factors influence the sound level reaching the listener's tympanic membrane: (a) the resonance characteristics of the external auditory canal and concha (Shaw, 1966); (b) the diffraction of the head and pinna (Weiner & Ross, 1946); and (c) the reflection and absorption characteristics of the listener's body. Although these factors are not accounted for when the substitution method of calibration is employed, the technique is appropriate for many audiologic investigations. Specifically, it provides a standard method of measurement for comparison among laboratories, measurements of stability of a particular environment, and a constant reference level with which to describe the behavior of human listeners in a particular acoustic environment. The differences in measures obtained by this method versus a real ear method (probe tube microphone) of calibration are usually less than 5 dB for frequencies below 2000 Hz. Complications arise for frequencies above 2000 Hz where the effects of ear canal resonance and head geometry become significant (Dirks, Morgan, & Wilson, 1976).
The technique of performing the calibration and measurement of the sound field also may influence the measures obtained. Use of a hand-held SLM or an octave-band analyzer is convenient and usually suitable for this type of measurement. If the instrument is hand-held, the body of the tester and the instrument case can affect the measured SPL, especially when the tester is close to the microphone. These effects are more pronounced when they are made close to the sound source in a room with boundaries that are highly absorptive. Although these effects are frequency-dependent, they can be ignored for frequencies below 300 Hz. This problem can be minimized by having the person performing the measurement hold the meter with arms extended while the sound from the sound source comes across the diaphragm of the SLM (Wells, 1979). The artifacts of the measurement procedure can be virtually eliminated by using a microphone on an extension cable with the measuring equipment and placing the observer some distance from the sound source and the microphone. In order to avoid body baffle effects when making measurements at high frequencies, the use of a microphone on an extension cable is particularly important (Wells, 1979).
The substitution method provides that the SPL of a test stimulus be specified at a “test point” in the sound field. The specification of the test point will depend on near/far field considerations and direct/reverberant field considerations. For most test environments the distance between the test point and the loudspeaker should be approximately 1 meter (Dirks et al., 1976) and must not be in close proximity to the walls of the sound enclosure or any other reflective surfaces. To specify the SPL using the substitution method, the diaphragm of a free field microphone is placed at a 0° incidence and the diaphragm of a pressure microphone is placed at a 90° incidence relative to the loudspeaker. The microphone height should be at the average height of the pinna of the listeners to be tested in that environment. Care should be taken to consider the relative differences in head position between children seated in small chairs and adults to be seated in higher chairs. In the unobstructed sound field, the SPL produced by the loudspeaker at positions of ±0.15 meters (5.9 inches) from the test point in both horizontal and vertical positions should deviate by no more than ±2 dB from the SPL at the test point. For specification of the SPL of frequency specific stimuli, the test signals are presented at a level sufficiently intense to exceed the ambient noise in the room. The SLM may be used in a linear mode or filtered appropriately for band limited signals. If filter settings are used, care must be taken to assure that the bandwidth of the signal does not exceed the bandwidth of the measuring system. All frequency specific test stimuli may then be specified in SPL.
Speech is a complex stimulus, varying over time in frequency and intensity. As a result, there are problems with sound field measurements of speech stimuli that are not associated with measurements of frequency specific constant-amplitude stimuli. The overall effects of the sound field on threshold are not as great as for frequency specific stimuli, but no standard exists defining the accepted method for calibrating any particular speech sample in the sound field. To explain, the current standard for audiometers (ANSI S3.6-1989) recommends that the “…SPL of a speech signal at the earphone is defined as the rms SPL…of a 1000 Hz signal adjusted so that the VU meter deflection produced by the 1000 Hz signal is equal to the average peak VU meter deflection produced by the speech signal.” However, the problems inherent in delivering a pure tone in the sound field (outlined in previous sections of this paper) obviously preclude the use of a 1000 Hz pure tone as a calibrating stimulus for speech. Additionally, the irregularities in the frequency response of loudspeakers make the use of any pure tone stimulus less than ideal as a calibrating stimulus for speech in the sound field. Alternatively, the suggestion has been made (Dirks, Morgan & Wilson, 1976; Dirks, Stream & Wilson, 1972; Tillman, Johnson & Olsen, 1966; Wilber, 1985) that a wide-band stimulus such as speech spectrum noise be introduced through the entire sound delivery system to determine the SPL developed at the test point (calibration position) in the sound field. The amplitude of the noise then is adjusted to produce a VU meter deflection that is equivalent to the average peaks of the speech signal. This procedure implies that the two signals are equated electrically via a VU meter. Because the frequency response of the loudspeaker affects the calibrating signal and the speech signal equivalently, the SPL of the speech spectrum noise will be a close approximation to the SPL of the speech signal.
Excessive ambient background noise can influence the accuracy of audiologic measures in the sound field in at least two ways. The background noises could render the test signal inaudible by direct masking. The listener's performance can also be influenced by distractions created by the intrusion of transient noises. Usually the test environment is selected so that ambient noise is relatively constant and the distraction problem is minimal.
Exclusion of all background noise is not necessary for auditory threshold measurements or measures of supra-threshold auditory function. The usual purpose for measuring ambient noise levels is to estimate the potential masking effect on thresholds. This is particularly important in measuring threshold sensitivity which is near normal in the sound field. Under sound field conditions, the person being tested is not afforded the attenuation of the ambient background noise provided by the earphone case and cushion. Maximum permissible ambient noise levels that allow the measurement of hearing threshold sensitivity in the sound field at levels as low as the reference equivalent threshold SPLs specified in the American National Standard S3.6-1989 have been published by the American National Standards Institute in ANSI S3.1-1977 (R1986). This document lists octave- and 1/3 octave-band levels that should produce no more than 1 dB of masking for standard monaural threshold hearing levels. The 1/3 octave and octave-band levels appropriate or sound field testing are presented in Table 2.
Ambient noise levels should be measured under the noisiest environmental conditions in which hearing tests may be conducted. Whenever possible, the following conditions should be observed: (a) the same number of people should be present both inside and outside the test room as during the routine testing activities, (b) air conditioning and ventilation equipment should be turned on if such equipment is to be operated during audiometric testing, (c) the noise measurement should be made at the time of day when background noise is at a maximum, and (d) additional noise measurements should be made during periods of intermittent noise such as those produced by typewriters, telephone bells, footsteps, etc. Occasionally, a noise which is relatively rare may intrude into the ambient test environment, (e.g., from aircraft flyover or a passing train). Testing should be interrupted when the offending noise is present (Appendix B, ANSI S3.1-1977).
The measurements of ambient noise in the sound field are made at the test point normally occupied during the test by the geometric center of the listener's head. Because ambient noise usually is diffuse, a random-incidence measuring microphone (sometimes called a “free field” microphone) is appropriate, and thus the orientation of the microphone diaphragm is irrelevant. However, when a noise is obviously coming from a particular direction, the microphone should be pointed toward the source, and the correction for normal incidence measurements for the microphone should be used (Peterson & Gross, 1972).
For the measurement of speech recognition thresholds (SRTs) in the sound field, it is assumed that any sound field that meets the permissible ambient noise levels shown in Table 2 is also adequate for speech threshold testing under the same conditions. For measuring SRTs, components above the 4000 Hz octave band need not be considered. Noise levels below 250 Hz should be evaluated by testing normally hearing listeners and noting if there is any effect on SRTs (ANSI S3.1-1977).
This section details suggested equivalent threshold SPLs (ETSPLs) for normal-hearing sensitivity in a sound field. Although the data from which the speech threshold levels have been derived were reported as the difference measurements between earphone (minimum audible pressure) (MAP) and sound field (minimum audible field) (MAF) (Dirks, Stream, & Wilson, 1972), the data have been converted to SPL. Morgan and coworkers (1979) suggested that difference measurements be employed as the most direct way of equating pure-tone measurements made under earphones with FM tone measurements obtained in the sound field. However, that procedure introduces a second source of variability into the measurement of threshold: the earphone measurement. Ideally, sound field calibration should not be dependent on any particular earphone standard; data reported in the future should be based on the independent measurement of threshold levels in the sound field without specific regard to the differences between earphone and sound field sensitivity. However, it is necessary to assure normal hearing of subjects using the recognized earphone standard. In the sections to follow sound field ETSPLs for speech and frequency specific stimuli are suggested.
Reports of earphone and sound field (MAP/MAF) differences for speech have been investigated and reported in several publications (Breakey & Davis, 1949; Dirks, Stream, & Wilson, 1972; Stream & Dirks, 1974; Tillman, Johnson, & Olsen, 1966). There is general consistency among the results regarding the differences between hearing threshold level for speech under earphones and in the sound field. However, that difference is dependent on the location of the loudspeaker in the sound field relative to the listener. For example, the report of Dirks, Stream, and Wilson (1974) reveals threshold differences for spondaic words between earphone and three loudspeaker locations: 0°: 3.5 dB; 45°: 7.5 dB; and 90°: 5.0 dB.
The methods for equating the speech calibration signal to the instantaneous rms-levels of speech stimuli and specifying the SPL of speech in the sound field were described earlier in the section on “Measurement of Test Signals”. Table 3 includes the suggested ETSPLs for speech in the sound field for three azimuth locations (0°, 45°, and 90°). The levels are based on the MAP/MAF differences reported by Dirks, Stream, and Wilson (1972) and Morgan, Dirks, and Bower (1979).
|Stimulus||0° a||45° b||90° a|
Walker, Dillon, and Byrne (1984) include suggested threshold sound pressure levels for a range of FM tones at 0° and 90° azimuth locations in the sound field. Table 3 includes the suggested ETSPLs for FM tones from 125 to 8000 Hz. Table 3 data at 45° are those reported by Morgan et al. (1979). Recall that the bandwidth characteristics of the Morgan et al. stimuli did not meet the bandwidth specifications suggested (Table 2) in the Walker et al. investigation. It should be noted that the work of Dillon and Walker (1982) reveals that the same SPL values should apply to NBN stimuli if the bandwidth of the NBN is comparable to the narrow or standard bandwidths suggested in Table 1.
The ETSPLs shown in Table 3 are suggested as interim values for the calibration of the sound field in order to express sound field thresholds with hearing threshold levels (HTLs) obtained under earphones. Using the Walker et al. (1984) values, Cox and McCormick (1987) compared thresholds obtained in the sound field to those measured under earphones for a group of normal listeners. Their data suggest that these ETSPLs are appropriate for the calibration of sound field warble tones. As such, the levels may be used as a reference for determination of deviation from normal hearing sensitivity in the sound field. These ETSPLs have not been accepted as a national or international standard. Consequently the values cannot be called reference ETSPLs. However, the ETSPLs reflect the most current data available.
Although there are no standards or guidelines for sound field testing, it is recognized that such testing is an integral part of audiologic evaluation. This paper has reviewed some of the problems in sound field testing, as well as possible solutions to those problems. This review may be summarized as follows:
The environment in which sound field testing is conducted is an integral part of the test procedure; thus, the ambient noise and reverberation characteristics of the test room must be known. The test room must have ambient noise levels below the level at which the test signals will occur.
The listener must be seated so that the SPL of the test signal is known at that listener's pinna. Thus, care must be taken to exclude anything between the ear of the listener and the loudspeaker, and the height of the loudspeaker must be appropriate for the listener. (Note: If the loudspeakers are raised or lowered, it may be necessary to recalibrate.) The near/far field and direct/reverberant field boundaries should be identified and the listener positioned between those two boundaries.
The acoustic properties of the test signal must be defined clearly. An FM signal is best for assessing threshold of hearing. The examiner should measure the SPL and verify the spectral characteristics of the signal. The frequency of calibration measurements should be identical to that used for earphones, generally once every 3 months.
Finally, it is important to understand the potential interaction between the test environment, the signal, and the listener when testing in the sound field. If the problems are understood and compensations are made it should be possible to obtain reliable and useful auditory information in the sound field.
American National Standards Institute. (1977). American national standard criteria for permissible ambient noise during audiometric testing (ANSI S3.1-1977 (R 1986)). New York: Acoustical Society of America.
American National Standards Institute. (1984). American national standard method for the measurement of the real ear attenuation of hearing protectors (ANSI S12.6-1964). New York: Acoustical Society of America.
American National Standards Institute. (1986). American national standard specification for octave-band and fractional-octave-band analog and digital filters (ANSI S1.11-1986). New York: Acoustical Society of America.
Dillon, H., & Walker, G. (1980). The perception by normal hearing persons of intensity fluctuations in narrow band stimuli and its implications for sound field calibration procedures. Australia Journal of Audiology, 2, 72–82.
Dirks, D. D., Morgan, D. E., & Wilson, R. H. (1976). Experimental audiology. In C. A. Smith & J. A. Vernon (Eds.), Handbook of auditory and vestibular research methods (pp. 517–531). Springfield, IL: Charles C Thomas.
Stephens, M. M., & Rintelmann, W. F. (1978). Influence of audiometric configuration on pure-tone, warble-tone, and narrow-band noise thresholds of adults with sensorineural hearing loss. Journal of the American Auditory Society, 3, 221–226.
Walker, G. (1988). Technical considerations for sound field audiometry. In R. Sandlin (Ed.), Handbook of hearing aid amplification, Vol. I, Theoretical and technical considerations (pp. 147–164). Boston, MA: College-Hill Press.
Ambient noise: The noise associated with a given environment, usually a composite of sounds from a multitude of sources near and far.
Amplitude modulation: The process by which the amplitude of a high-frequency signal (carrier) is controlled by a low-frequency signal (modulator).
Anechoic room: A room in which the boundaries effectively absorb all incidence sound, thus simulating a free field.
Angle of incidence: The angle between the direction of travel of a sound wave and the microphone diaphragm.
Antinode: The point, line, or surface in a standing wave where some specified characteristic of the wave has maximum amplitude.
Baffle: A shielding structure used to increase the effective length of the transmission between two points in an acoustic system (e.g., between the front and back of a loudspeaker).
Carrier frequency: The center or nominal frequency of a complex signal which may be modulated by a variety of waveforms (e.g., square wave, triangular wave, sine wave, or ramp).
Damped wave train: An exponentially decaying sinusoidal tone burst presented at a specified repetition rate.
Diffraction: Phenomenon by which the direction of a sound wave is changed by an obstacle (e.g., a human body) in the medium.
Diffuse sound field: A field in which reflected waves arriving equally from all directions combine to produce uniform sound pressure at all points within the field.
Digital synthesis: A system of constructing signals through mathematical description rather than through measurable physical construction (e.g., computer development).
Directionality (Directivity factor): The ratio of the sound pressure squared at a fixed distance and direction from a transducer to the mean square sound pressure at the same distance averaged over all directions from the transducer.
Direct field: That portion of a sound field (usually closest to the source) where the SPL of the free field exceeds the SPL of the diffuse field.
Equivalent threshold sound pressure level (ETSPL): The sound pressure level set up by a transducer at a specific frequency in a specified coupler when the transducer is activated by that voltage which, with the transducer applied to the ear concerned with a specified force, would correspond with the threshold of hearing.
Efficiency: The ratio of the useful output of a device to its total input.
Far field: The soundfield at a distance from the source where the particle velocity is primarily in the direction of propagation of the sound and the acoustic intensity is proportional to the sound pressure squared (see Inverse Square Law).
Field uniformity: Where the sound level within a given space is equivalent at all points within that space.
Free field: A field in a homogenous, isotropic medium in which the effect of boundaries are negligible.
Free field microphone: A microphone whose output is independent of the angle of incidence of its face to the sound source.
Frequency deviation: In frequency modulation, the difference between the modulation frequency and the center frequency.
Frequency modulation: A stimulus in which a modulation signal controls the frequency of a carder (usually a sine wave).
Frequency response: The reference pressure response presented as a function of frequency.
Inverse square law: A rule that in an anechoic chamber governs sound radiation in an acoustic far field of a sound source. For each doubling of distance from the source, the SPL is halved. The change is usually expressed as a decrease in SPL of 6 dB.
Loop (antinode): A point, line, or surface in a standing wave where some characteristic of the wave field has maximum amplitude.
Minimal audible field (MAF): The minimum audible field is the SPL of a tone at the threshold of audibility measured in a free soundfield. The SPL is measured in the absence of the listener in the unobstructed soundfield.
Minimal audible pressure (MAP): The minimal audible pressure is the SPL of a tone at the threshold of audibility when the signal is presented via an earphone. The SPL may be measured at the tympanic membrane or in an acoustic coupler designed to reflect the transfer characteristics of the external ear.
Modulation rate: The speed at which the signal is changed (i.e., number of times the signal is changed per second).
Narrow band noise: A noise that has been filtered or constructed to have equal amplitude in a specified band of frequencies (e.g., narrow band masking noise whose bandwidth is defined dependent on center frequency).
Near field: The acoustic near field of a sound source is the field close to the sound source where the instantaneous sound pressure and particle velocity are out of phase.
Node: A point, a line, or surface in a standing wave where some characteristic of the wave field has essentially zero amplitude.
Null: That point at which a node is reached.
Pressure microphone: A microphone for which the electric output corresponds to the instantaneous sound pressure of the incoming wave.
Pulse train: A series of periodic signals usually consisting of rectangular waves.
Pure tone: A signal in which the instantaneous sound pressure varies as a simple sinusoidal function of time.
Rejection rate: Attenuation of a signal in dB/octave of a signal outside its nominal frequency band.
Reverberant field: The portion of a sound field where the SPL of the diffuse field exceeds the SPL of the free field.
Reverberation: The persistence of sound in an enclosed space as a result of repeated reflections from surfaces following the termination of the incidence signal.
Root mean square (rms): The effective sound pressure at a point, defined as the square root of the mean value of the squares of the instantaneous values of the quantity over a time interval.
Sound field: A field (or defined space) containing sound waves.
Spectrum: A description of a complex signal in terms of the frequency, amplitude, and phase of each individual component.
Standing wave: A periodic wave having a fixed distribution in space which is the result of interference of progressive and reflected waves of the same frequency. Standing waves are characterized by the existence of nodes or partial nodes and antinodes that are fixed in space.
Transducer: A device capable of changing one form of signal energy (e.g., electric) into another form of energy (e.g., acoustic).
Waveform: A description of the shape of an acoustic signal (e.g., square waveform; sinusoidal waveform).
|Stimulus||0° a||45° b||90° a|
Index terms: acoustics, assessment
Reference this material as: American Speech-Language-Hearing Association. (1991). Sound field measurement tutorial [Relevant Paper]. Available from www.asha.org/policy.
© Copyright 1991 American Speech-Language-Hearing Association. All rights reserved.
Disclaimer: The American Speech-Language-Hearing Association disclaims any liability to any party for the accuracy, completeness, or availability of these documents, or for any damages arising out of the use of the documents and any information they contain.