Audibility, Distortion, QSIN, HINT, and SNR Loss
For the last century, hearing professionals have approached hearing loss with respect to making sounds louder to compensate for a lack of audibility (i.e., audibility decreases as hearing thresholds increase). Grant (2013) notes that making things louder is more-or-less the "low hanging fruit" with regard to signal processing for people with hearing loss. He reports that the long-term goal of hearing aid fittings is to incorporate signal processing algorithms that "reverse engineer" distortions in sound, similar to how eyeglasses make fuzzy images clear. Unfortunately, simple refractive solutions are not possible with hearing loss, as sound distortions involve multiple factors such as sensorineural hearing loss (SNHL), loss of frequency and temporal resolution, signal-to-noise ratio (SNR) loss and multiple cognitive issues (speed of processing, working memory, attention and more).
Grant reports that it is easy to characterize a signal reaching a patient’s ear, but it’s much more difficult to send a signal through an impaired cochlea and then characterize the signal reaching the brain! Of course, the cognitive ability of the brain (processing ability, processing speed, knowledge, vision, working memory, attention, and more) interacts with (and impacts) the original signal to create a unique auditory percept.
Grant and Walden (2013) report that traditional measures from typical audiometric diagnostic tests (pure-tone thresholds, word recognition scores in quiet, etc.) are of little consequence in predicting the ability of people with hearing loss to understand speech in noise (with or without amplification). They report "none of these tests administered in quiet reliably predicts a patient’s speech recognition performance under noisy conditions…."
Grant and Walden evaluated 27 people with hearing loss and 4 people with normal hearing using the fixed presentation level QuickSin (QSIN) and the adaptive Hearing in Noise Test (HINT). They report that the average HINT presentation level (near the SRT) was 68 dB SPL, whereas the average QSIN presentation level was 85 dB SPL. As such, the QSIN often allowed more audibility for hearing impaired listeners, than did the HINT. Specifically, when administered as prescribed, the HINT is less audible than the QSIN. However, this same factor (increased audibility) may lead to increased distortions.
The authors note that the different background noises appeared to not be significant issues (HINT uses speech-shaped steady state noise, QSIN uses multi-talker speech babble). Grant and Walden report that the QSIN and HINT were significantly correlated with each other (r = 0.8), indicating the two tests assess similar (although not identical) processes, as regards speech in noise. The authors note after audibility has been addressed, suprathreshold measures of distortion are (likely) useful with regard to understanding individual differences with respect to speech recognition in noise.
For More Information, References, and Recommendations
Beck DL, Nilsson M. (2013) Speech-in-Noise Testing: A Pragmatic Addendum to Hearing Aid Fittings Hearing Review May(16):37.
Grant KW. (2013) Guest Editorial – Auditory Research at the Walter Reed National Military Medical Center. Journal of the American Academy of Audiology (JAAA), 24(4).
Grant KW, Walden TC. (2013) Understanding Excessive SNR Loss in Hearing Impaired Listeners. Journal of the American Academy of Audiology (JAAA) 24(4):258-273.