A5 Loss of Speech Intelligibility in Noise

Topic

When someone speaks in a noisy environment, the speech signal does not reach the ear in isolation; it arrives mixed—within the same sound wave—with the competing background noise. Whether that speech signal is intelligible depends not merely on its absolute loudness, but on how well it stands out against the accompanying background noise—a relationship known as the acoustic signal-to-noise ratio. A favorable acoustic signal-to-noise ratio (a strong speech signal relative to low background noise) facilitates comprehension, whereas an unfavorable ratio (a weak speech signal relative to high background noise) hinders it, even before any issues involving the listener's ear or brain come into play.

The signal-to-noise ratio a listener can actually utilize depends not only on the acoustic mixture itself but also on how effectively their peripheral ear captures the speech signal within that mixture. When peripheral hearing loss reduces the audibility of the speech signal or degrades the resolution with which the inner ear encodes its fine spectral and temporal details, the listener loses the ability to capitalize on many of the subtle differences distinguishing the speech signal from the background noise—even if the acoustic signal-to-noise ratio of the original mixture remains unchanged. Consequently, for that listener, peripheral hearing loss lowers the effective signal-to-noise ratio below the actual acoustic signal-to-noise ratio; they require a mixture that is proportionally more favorable to the speech signal than what a listener without peripheral hearing loss would need to achieve the same level of understanding.

The effective signal-to-noise ratio depends not only on what the peripheral ear manages to capture but also on how well the listener's central processing separates the speech signal from the background noise within the incoming sound. When central processing is impaired, the listener loses some ability to track the fine temporal and spectral details that would allow them to distinguish the speech signal from background noise—even when the peripheral ear delivers these details intact. Thus, impaired central processing further reduces the effective signal-to-noise ratio, independently of any reduction already caused by peripheral hearing loss. When both factors coexist in the same listener, their effects on the effective signal-to-noise ratio are additive, resulting in an effective signal-to-noise ratio that is considerably lower than the original acoustic signal-to-noise ratio of the mixture.

Since peripheral hearing loss and impaired central processing affect the effective signal-to-noise ratio in different ways, this model quantifies their combined impact by comparing two measurements from the same listener: pure-tone recognition—a task requiring only the detection of a simple sound in quiet, which primarily reflects the state of the peripheral ear—and speech-in-noise recognition, a task that additionally requires separating the speech signal from background noise using central processing. When pure-tone recognition is near-normal but speech-in-noise recognition is disproportionately poor, the discrepancy between the two measurements indicates that impaired central processing is independently reducing the effective signal-to-noise ratio, beyond what would be explained by peripheral hearing loss alone.

Because the ultimate difficulty is measured by the listener's ability to track the speech signal amidst background noise—the scenario that introduced this model—corrections target each of the two factors that reduce the effective signal-to-noise ratio. A hearing aid with noise processing primarily addresses the peripheral aspect of the problem: it attenuates background noise and enhances the speech signal before both reach the ear, thereby improving the acoustic signal-to-noise ratio actually received by an already impaired peripheral ear. Auditory training, on the other hand, targets the central component: it exercises impaired central processing to restore the efficient extraction of the fine temporal and spectral details that distinguish the speech signal from background noise, thereby increasing the effective signal-to-noise ratio even without altering the acoustic mix reaching the ear.

ID:4245