Dr. Brennan, left, is a staff scientist and audiologist at Boys Town National Research Hospital in Omaha, NE, where Dr. McCreery, right, is also a staff scientist, as well as associate director of audiology.
Children listen in a range of acoustic environments, some of which are less common for adults, such as the cradle position for infants and the classroom environment for older children.
The level of the talker's voice that reaches the child will vary across these listening situations, and audibility might differ as well, even with proper amplification.
While our standard measures of audibility do not account for the listening situations typical for children, the Situational Hearing Aid Response Profile (SHARP) does.
New updates to this computer program enable accurate predictions of hearing aid output levels and allow the clinician to visualize audibility for non-lowered and frequency-lowered signals.
Audibility can be estimated with the speech intelligibility index (SII), and a modified version of the SII is available for use with nonlinear frequency compression (NFC).
MANUAL DATA ENTRY NOT REQUIRED
SHARP was originally developed to estimate audibility for 13 different listening situations. (Stelmachowicz PG. Situational hearing aid response profile. In Bess FH, Gravel JS, Tharpe AM, eds. Amplification for Children with Auditory Deficits. Nashville, TN: Bill Wilkerson Center Press; 1996.) The program could be used for unaided listening or for linear and single-channel compression amplification.
To estimate audibility, the clinician manually entered hearing thresholds and hearing aid electroacoustic characteristics from the verification process. This process was a limiting factor for busy clinicians who did not have the time or desire to input their patient's verification data into a computer program.
Addressing this limitation, SHARP now allows the user to import the electroacoustic characteristics of the hearing aid from the Audioscan Verifit, RM500SL, or Axiom verification systems.
Audibility is then estimated for each listening situation based on linear regression. (Brennan M, McCreery R, Lewis D, et al. The situational hearing aid response profile: an update. Poster presented at: the 40th Annual AAS Science and Technology Conference of the American Auditory Society; March 7-9, 2013; Scottsdale, AZ. https://aas.memberclicks.net/assets/docs/aas_2013_poster_abstracts.pdf?mcid_token=36108aed-843a-477d-a9b9-c9469c6b1707)
Until recently, SHARP had not been updated to reflect recent advances in hearing aid signal processing, such as multichannel amplitude compression and nonlinear frequency compression. While the use of NFC has increased in children, estimating audibility with the approach remains challenging.
Nonlinear frequency compression splits the input signal by the start frequency. Sounds above the start frequency are compressed to a lower frequency, reducing the frequency difference between sounds. This compression has the potential to increase audibility for higher-frequency sounds.
The amount of compression is dictated by the compression ratio. The higher the compression ratio, the tighter the squeeze of high-frequency sounds into a smaller region.
SPEECH INTELLIGIBILITY INDEX
We developed a method for displaying audibility of different listening situations with SHARP.
Mapping of the input frequencies to the output frequencies was derived from the SoundRecover Fitting Assistant (Semin Hear 2013;34:86-109 https://www.thieme-connect.com/ejournals/abstract/10.1055/s-0033-1341346) and validated internally using measurements from Phonak and Unitron products. (Brennan M, McCreery R, Lewis D, Kopun J, Stelmachowicz P. Incorporating measures of nonlinear frequency compression into the situational hearing aid response profile (SHARP). Poster presented at: A Sound Foundation Through Early Amplification; Dec. 8-11, 2013; Chicago, IL. http://phonakchicago2013.com/web/poster_marc_brennan.php)
The SHARP program is depicted in figure 1. To get this figure, we first performed real-ear measurements in a child for speech inputs of 50 and 60 dB SPL and an 85-dB SPL swept pure tone. That data was then imported into SHARP to estimate audibility.
Figure 1 represents the amount of audibility we would expect when a classroom teacher is four meters from the listener. The shaded region shows the audible portion of the speech spectrum, and the unshaded region shows the inaudible portion.
Different listening situations can be selected on the right side. Figure 2 shows estimated audibility for conversational speech at four meters from the same listener. Notice that audibility is lower for conversational speech than it is for a classroom teacher's speech because of the teacher's greater vocal effort.
Audibility is also represented in the lower left-hand corner by the speech intelligibility index. (ANSI/ASA S3.5-1997. Methods for the Calculation of the Speech Intelligibility Index. New York, NY: American National Standards Institute.)
The speech intelligibility index ranges from zero to 100 percent, with zero meaning that none of speech is audible and 100 that all of speech is audible.
It is calculated by computing the sensation level for different frequency bands and then multiplying the sensation level by an importance function for that frequency band. The importance function represents the amount of speech information contained within each frequency band.
When a classroom teacher is four meters away from the child described in figure 1, the SII is 86 percent, suggesting that 86 percent of speech is audible for this listener. For conversational speech, the SII is less—73 percent for this listener.
NONLINEAR FREQUENCY COMPRESSION
To display nonlinear frequency compression, the clinician checks the “Freq Comp” box and then enters the start frequency (FCC) and compression ratio (FCR), as shown in figures 3 and 4.
Here, the start frequency is 2 kHz, and the compression ratio is 2. Frequencies greater than 1.5 kHz—the lowest start frequency available with Phonak and Unitron products—are plotted in orange.
The diamonds are closer together in figure 3 than in figure 1, demonstrating the effect of frequency compression. The markers that represent the highest three frequencies—5, 6.3, and 8 kHz—are only audible with nonlinear frequency compression (figure 3).
SHARP also provides an SII measurement with nonlinear frequency compression. In this example, the speech intelligibility index when a classroom teacher is four meters from the listener is 12 percentage points higher with NFC than without it—98 percent (figure 3) and 86 percent (figure 1), respectively.
With nonlinear frequency compression, the speech intelligibility index is calculated by computing the importance function based on the input frequency and the sensation level based on the output frequency.
Notice that this represents the proportion of speech that is audible but not necessarily the amount of speech that the listener is able to use effectively for speech understanding.
That is, although you might see an increase in the speech intelligibility index with nonlinear frequency compression, the listener might not achieve an equivalent improvement in speech recognition.
However, our own results suggest that this method of estimating audibility is accurate, on average, at predicting performance with nonlinear frequency compression. (McCreery RW, Alexander J, Brennan MA, Hoover B, Kopun J, Stelmachowicz PG [in press]. The influence of audibility on speech recognition with nonlinear frequency compression for children and adults with hearing loss. Ear Hear.)
Copies of SHARP can be obtained by e-mailing firstname.lastname@example.org.© 2014 by Lippincott Williams & Wilkins, Inc.