Decibel scales and Calibration

Caution using A-weighted decimal scales  -  Decibels and hearing level (HL)

Calibrating an Audiometer in dB HL  -  Caution calibrating sound fields

Audiologists need to know the difference between dB(A), and dB HL and dB SPL and how each are used in calibration.  It is necessary to be aware of the limitations of the dB(A) scale when examining threshold measures and also recognize the differences between dB HL and dB(A) when comparing data.     

As we all know, the decibel is a logarithmic unit that is used to express the relative magnitude of two quantities (the ratio of two quantities).  However, decibels start to become confusing when we introduce the notion of SPL vs HL vs weighting scales.

For instance, when we say a whisper is 30 dB SPL, it is 30 dB in regards to an accepted reference value.  dB SPL is the unit of pressure referenced relative to 20 mPa.  

In terms of human loudness perception, the problem with a dB SPL scale is that the loudness of a sound is not necessarily related to it's dB in SPL.  Therefore, the dB(A) scale was developed to account for the subjective loudness of sounds (Hassall & Zaveri, 1979).

Human loudness perception depends on the sound intensity at the eardrum (as well as its frequency, bandwidth and duration).  The sound intensity outside the ear does not necessarily reflect the sound intensity at the tympanic membrane.  The head, torso as well as the ear canal resonance effects, affects the sound intensity at the tympanic membrane.  These effects are referred to as the field-to-drum transform characteristics.  Because of these transformation characteristics, the SPL at the tympanic membrane differs from that measured in the sound field, and these differences vary with frequency.  The SPL that corresponds to the threshold of hearing at the eardrum is referred to as the minimal audible pressure or MAP.      

The most common reference for intensity in clinical Audiology is dB HL referenced to either ANSI or ISO standards.  These standards were derived from behavioral threshold measures on very large groups of normal hearing subjects.  The dB HL scale therefore, is a general term for a scale that reflect the intensity of sounds at different frequencies and measures them with respect to the threshold of hearing (0 dB HL being threshold).

Caution using A-weighing decibel scale:

The dB(A) scale was developed to try and account for the subjective loudness of sounds; that is to try and account for the fact that the loudness of a sound perceived is not necessarily directly related to its power in dB SPL.  The sound level pressure that exists at the tympanic membrane is different from that measured in a sound field and this difference varies with frequency.  The transform characteristic of the sound field to the ear drum affects the perceived loudness of a sound and the dB(A) weighted scale attempts to account for this.

Thresholds obtained using a dB(A) scale present several limitations that an Audiologist needs to be aware of.  The dB(A) scale was developed for a subject sitting in a sound field that involves a frontal incidence sound wave in a room with no reflections (Beynon 1993).  If reflections are present in the sound field, or the direction of the sound source is altered somehow (no longer frontal incidence), then these alterations in the sound presentation results in changes in the sound intensity at the eardrum.  In essence, the transformation characteristics that the dB(A) scale was accounting for, is changed.     

When comparing sound in a free field and diffuse sound field, sound in a diffuse field will be more intense at the listener’s eardrum than in a free field.  If a dB(A) scale is used in a diffuse field, then the level would be incorrectly measured as being lower than it actually is.  

Another cautionary note is that the zero point on a dB(A) scale is an arbitrary point that is not related to the threshold of hearing.  The dB(A) scale is based on a 1 kHz tone at 40 dB SPL (ISO 131, 1979).  This does not mean that at 40 dB SPL, the sound is 40 dB above the minimal audible pressure (MAP).  40 dB(A) is not 40 dB above threshold.  0 dB(A) reflects the perceived loudness of a 1 kHz tone at 0 dB SPL, whereas the MAP is often 5 dB above this. 

For Audiologists this is an important fact to know because when you are measuring thresholds with dB(A), thresholds will be higher at all frequencies except at 3-4 kHz compared with thresholds done using a dB HL scale.  As a result the dB(A) scale is not applicable in situations that vary significantly from a wave of sound coming from a source directly in front.  Therefore, the dB(A) scale is not suitable for many audiological situations.  dB(A) scales are however used in sound field testing, but since the test rooms used are not truly anechoic, Audiologists must recognize the inherent error that may occur in the measurements using this scale. 

As Beynon 1993 states, "Use of the dB(A) scale in clinical measurements is therefore flawed...it is expressly designed for use in a true free-field which is rarely encountered in clinical testing, and its arbitrary choice of 40 dB SPL as a reference results in the dB(A) scale not measuring hearing relative to threshold."    

A common misconception is that the dB(A) scale is a good approximation to the dB HL scale that is suitable for sound field testing.  Since the dB(A) scale uses a set of figures derived from a free field situation, it does not overcome the problems of finding the appropriate RETSPL values for a test room.  For a detail explanation on finding RETSPLs for a test room and the problems associated with it, click here.  

Decibels and hearing level (HL):

Decibels in hearing level (HL) reflect the intensity of a sound at different frequencies with respect to the listener’s threshold level.  In order to give accurate values in dB HL, the variation in sensitivity at the tympanic membrane and the transform characteristics present must be accounted for.  The transformation characteristics vary depending on how the sound is present and so different correction factors must be found for each method of presentation.  By knowing the transformation characteristics, (difference in SPL for sound at the transducer and at tympanic membrane for each test frequency), we could add them to the MAP (minimum audible pressure) to find the output level that should correspond to threshold.  This value is different for everyone because no two people have exactly the same ear canal in terms of shape, size and impedance characteristics and an exact measure would require finding correction factors specific to each patient.        

Calibrating an Audiometer in dB HL: 

To calibrate a set of earphones in dB HL for an audiometer, the following should be done.  First, the transformation characteristics for an earphone need to be known (i.e. the difference between the SPL next to the earphone and that at the eardrum for each frequency).  Knowing the transform values and subtracting them from the MAP, allows you to determine the output level which corresponds to threshold.   For example, if the MAP at a certain frequency is 30 dB SPL and the transform characteristic at that frequency added an additional 5 dB of intensity at the eardrum, then you know that an output of 25 dB (30 - 5) at the earphone will be at threshold at the eardrum.  The question now is, how do we do this?   

As mentioned earlier, the hearing level scale (dB HL) is developed by finding the threshold of a large number of otologically normal subjects.  When an audiometer is calibrated to measure hearing level, the dial reading on the audiometer for the average threshold at each frequency will read 0 dB HL.  This biological calibration method sets the output levels at 0 dB HL, which corresponds to threshold according to the standardized HL values.    

Next, we want to record the audiometer output in a coupler.  Measuring and recording the audiometer output SPL created in a coupler with the audiometer dials set to 0 dB HL allows you to calibration other audiometers by using these recorded values.  To calibrate another audiometer, place the earphone on a coupler and vary the output levels on the audiometer until they are the same as the recorded values. 

The coupler value does not reflect the actual levels obtained next to the eardrum because each ear is characterized by different impedances and volumes.  It does however provide a reference point to obtain the same output level as originally found. 

If a different earphone or coupler is used, then the level generated  in the coupler may be different that what was originally found.  Therefore, when calibrating an audiometer, one must ensure that the earphones and earphone cushions are the same or perform the same as the ones originally used.  The threshold values obtained using this method are called reference equivalent threshold sound pressure levels (RETSPL).   

Caution calibrating the sound field:

When using dB(A) scales to measure hearing thresholds, it is important not to compare it to data in dB HL because they are not the same.  This is especially important when looking at data obtained in the sound field.  The procedure of calibrating a sound field such as a clinical sound booth is essentially the same as calibrating an audiometer, however, sound field calibration is inherently prone to more errors.  For example, when calibrating a sound booth, the threshold data of a large otologically normal group of subjects needs to be obtained.  The output level used to achieve threshold measures needs to be recorded (just like when using the coupler, except instead of a coupler, we have a sound booth).  The output levels are measured using a sound level meter placed at the location where the center of the subject’s head is located.  According to Beynon (1993), this procedure will take into account the transform characteristics and allow for recalibration using reference levels in much the same way as for headphone calibration.  Unlike headphone calibration, transform characteristic in sound field calibration depends on the type of sound booth and it’s physical properties, speaker location, as well as the subject’s physical position. 

In essence, an individual set of reference values have to be found for each individual test set-up.  Rather than having to go through this entire process again to calibrate a sound field, a set of reference values for sound-field testing for a typical audiometric test booth have been suggested by Walker et al (1994).  A cautionary note on using these values is that a range of +/- 10 dB has to be applied to sound-field results.  Therefore, an Audiologist using these figures must be aware of the possible errors involved.  Retesting does not help eliminate errors because the errors arise from both equipment variation and variation in testing.                

There are currently many methods used in reporting sound-field results.  Sound-fields calibrated in dB(A), or not calibrated at all may be recorded on audiograms and compared to earphone measurements in dB HL.  Audiologists need to know that when you compare values in dB(A) to dB HL (i.e. treat a dB(A) scale as if it were a dB HL scale calibrated for that sound field), substantial errors are possible.  Measurements in dB(A) marked on audiograms are very misleading and should be clearly marked as being on a different scale to the rest of the graph.  Also, comparing results in dB(A) that have been obtained in different testing environments can lead to confusion.  According to Beynon (1993), differences may be present that are due to the different sound field characteristics present rather than as a result of any changed in hearing sensitivity. 

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