Sound Field Calibration using Warble Tones

Sound booths used in sound field testing need to be calibrated so that thresholds measurements correspond to those values obtained using earphones.  Audiologist need to be aware that there are no standard RETSPL values used to calibrate the test booths but there are recommended stimuli and procedures used in calibration.  Although these values do allow for good sound field threshold measurements, Audiologists need to be aware that they may also contribute to measurement variability.  

The problem of standing waves in sound field testing can be reduced by using FM warble tones rather than pure tones.  Standardized RETSPL values are not currently available although several studies have presented a recommended procedure and set of stimuli used in calibration and conduction of sound field tests.  Walker et al. (1984) presented what they believe to be a good set of RETSPL for calibrating sound fields. 

Despite published reports of recommended stimuli (Walker et al 1984, Morgan et al 1979), different manufacturers of audiometers produce devices with slightly different FM warble tones.  In addition, sound booths and sound fields found in clinical testing are rarely the same (acoustically and architecturally) across clinics, which contributes to threshold measurement variability (Staab et al 1972).

Cox et al. (1987) conducted a study using Walker et al.’s (1984) RETSPL values across different sound fields and audiometers and found that the sound field thresholds were essentially the same as those obtained through earphones.  They also discovered that threshold variability due to asymmetric sound field characteristics and slightly different FM warble tones from different instruments can account for a small portion of statistically significant discrepancies between sound field threshold and corresponding earphone values.  By themselves however, the magnitude of these discrepancies may not be clinically significant but Audiologists need to be aware that they can contribute in an additive manner to the over all level of variability in a test measure. 

Sound fields thresholds are typically measured in audiometric test booths with a semi reverberant rather than anechoic characteristic.  In a semi reverberant environment, standing waves can be a problem and calibrating the sound field using pure tones may not be possible.  Pure tone stimuli are not recommended for sound field testing.  When using pure tone stimuli, the sound pressure level in a typical test room varies with distance from the loudspeaker.  Consider the following graph from Walker et al 1984.

  graph from Walker et al 1984

In this graph, the sound pressure level of a 1350 Hz pure tone stimulus is plotted as a function of distance from the speaker.  For distances up to 0.6 meters but beyond the near field, the SPL is mostly influenced by the sound coming directly from the speaker.  The sound varies approximately according to the inverse square law.  When the sound signal exceeds 0.6 dB from the speaker, the sound field can be said to have entered a reverberant field, and the SPL begins to be influenced by reflected sounds in addition to the direct sounds.  When this happens, standing waves can occur.  The graph above illustrates this interaction of direct and reflected sound as numerous peaks and valleys.

Audiologists need to know the impact of using pure tone stimuli in a reverberant field is that small movements from the test subject could result in significant variations in the SPL at the eardrum (Walker et al 1984).  The location and magnitude of the peaks and troughs in the above graph are frequency dependent, therefore, test results between audiometer calibration may vary significantly.  Even small changes in frequency, especially in very reverberant rooms, will can cause significant SPL variation at the eardrum using pure tone stimuli.         

The above problem with standing waves, does not occur if the subject is placed within 0.6 m from the speaker.  Placing the subject within this narrow range within the speaker is not recommended either since a) SPL changes rapidly with variations in distance in the direct sound field; b) it may be impractical place the subject so close.    

To help over come the effects of standing waves, frequency modulated (FM) warble tones can be used in calibration.  FM warble tones are less affected by standing waves in the sound field compared to pure tone stimuli because FM tones have a larger bandwidth.

An Audiologist wanting to measure thresholds in a semi reverberant field to reflect hearing loss, can calibrate the sound field using a biological method.  The problem with using this method is that ambient noise levels in the test environment may be high enough to mask the sound field threshold for sensitive normal listeners.  Audiologists need to know that most test facilities are not equipped to deal with this situation.  In addition, Audiologists need to be aware that audiometers are often nonlinear at low levels, which can lead to inaccurate measurements for normal people.  Managing both these issue is a time consuming process and unfortunately leads to infrequent calibration checks and a reliance on small data sets to derive calibration values.

Given these points, a standard RETSPL for warble tone sound field testing seems warranted.  The problem with establishing this standard is that there is a lot of variability in the FM warble tones across different audiometers.  Audiologists need to be aware that the modulation bandwidth, modulation waveform, and modulation rate differ considerably across instruments and these factors are often not described in the manufacturer specifications.  Another major problem is that a standard RETSPL would require a standard testing environment.  That is, the room size and acoustic characteristics must be the same for all sound booths for a standardized RETSPL to be used. 

A properly calibrated sound field would ideally provide hearing threshold levels that correspond to hearing threshold levels measured through earphones.  Cox et al. (1987) conducted a study to try and accomplish just this.   

Despite the obstacles in establishing RETSPLs for warble tones, Walker et al. (1984) introduced a set of recommended stimuli and procedure for these values.  To demonstrate the possible measurement variability that could arise from using Walker et al.’s (1984) proposed RETSLP values, Cox et al. (1987) conducted a study using two different sound fields and two different audiometers.  They found that in both sound field conditions, the mean sound field thresholds were significantly different from the earphone condition in 17% of the comparisons.  However the magnitude of most of these observed differences is clinically insignificant with the largest observed difference being 6.9 dB.  

They also pointed out that the Saico audiometer used in their study seemed to be slightly more detectable than the Grason-Stadler one.  An Audiologist my be interested in knowing that the reason for this difference may be due to a finding by Barry and Resnick (1978).  They found that instruments producing lower modulation rates results in slightly lower thresholds at all test frequencies regardless of the modulation waveform or bandwidth.  The mean difference however, between the two FM tones was only 3 dB, which by itself is clinically insignificant.   

Audiologist should be aware that the sound booth in which testing is performed might contribute to idiosyncratic sound field effects.  Cox et al (1987) and Morgan et al. (1979) found unpredictable and sometimes asymmetric threshold differences in sound field tests.    

What was demonstrated in this study demonstrates that you can use the RETSPLs suggested by Walker et al. (1984) even though the audiometer you have may generate FM warble tones with different modulation parameters than those used by Walker et al (1984).  The graph below is good summary of the result obtained from Cox et al. (1987).  The solid black bars represent the difference between the measured threshold value of sound field A and the measured earphone threshold.  The striped bars represent the difference between the measured threshold of sound field B and the measured earphone threshold.  Only 1% of the values obtained in both sound field conditions exceeded 10 dB of those measured using earphones (threshold difference 15 dB).   

   figure from Cox et al. 1997

When the sound field-earphone threshold differences are represented using 5 dB increments, the differences between them did not prove to be different than those observed in repeated clinical tests using earphones alone.

The Cox et al. (1987) study indicated that when the sound field and earphone calibration are perfectly performed, their data suggest that the normal variability of thresholds would result in a difference which would exceed 10 dB only 5.5% of the time with a test-retest variability which would exceed 10 dB 2% of the time.  The main point this study conveys is that essentially, all the sound field thresholds were within 10 dB of the corresponding earphone values.  Audiologist need to be aware that the variability that does occur in sound field testing is partly due to minor, probably predictable differences in detectability of warble tones with varying modulation parameter values and partly due to unpredictable, perhaps asymmetric, acoustic effects related to the specific sound field used.  Although the magnitude of the sound field-earphone difference is not likely to be clinically significant, an Audiologist should bear in mind that the variability can contribute to other areas of miscalibration in an additive manner that may result in clinically significant differences.       

Note:  This site is best viewed using IE 4.X or higher under 800x600 resolution