CI Atlas · When Hearing Aids Aren't Enough · Module 09

9Audibility vs distortion

The previous modules each described one distortion of the damaged cochlea. This one pulls them together into a single accounting. Aided speech understanding can be thought of as the sum of two parts: how much of the speech the listener can actually hear — audibility — and how much the damaged cochlea then garbles — distortion. For mild loss, audibility is almost the whole story, and audibility-based predictors like the Speech Intelligibility Index work well. But as loss deepens, the distortion penalty grows, and a point comes where the speech is fully audible yet still poorly understood, so making it louder adds nothing. That widening gap between audible and intelligible is the quantitative heart of why amplification eventually fails — and the signal to consider an implant.

TTwo components of understanding

It helps to split aided speech understanding into two parts: an audibility component (how much of the speech spectrum is above threshold and within the dynamic range) and a distortion component (how badly the cochlea garbles what is audible). A hearing aid can work hard on the first; it can do almost nothing about the second.

Models like the Speech Intelligibility Index predict understanding from audibility — and for mild loss they work well. But audibility has a ceiling, and the damaged cochlea subtracts a distortion penalty (recruitment, blurred resolution, dead regions) that audibility-based models do not capture. As loss deepens, you can still make speech audible, yet the penalty grows until making it louder buys almost nothing. That gap between audible and intelligible is what sends a patient toward an implant. Schematic.

CWhere audibility models break

Audibility can be quantified — the Speech Intelligibility Index (SII) weights how much usable speech information is audible — and for listeners with mild loss it predicts performance well. But the SII assumes that audible speech is usable speech. In greater losses that assumption fails: two ears with the same audibility can score very differently because their suprathreshold distortion differs, and audibility explains progressively less of the variance.[2007]

CThe growing distortion penalty

As loss deepens, the distortion penalty — recruitment, broadened filters, temporal smearing, dead regions — grows. Amplification can keep speech audible almost indefinitely, but the achievable score is capped by distortion and falls as the penalty rises. Eventually the curve goes flat: the speech is fully audible, yet making it louder yields no further understanding. That plateau is the practical ceiling of the hearing aid.

CWhen the fix becomes the problem

There is a final twist. Some of the tools used to maximise audibility can add their own distortion. Wide dynamic range compression, which helps fit speech into the narrowed range, can blur the consonant–vowel intensity differences that carry meaning, and may worsen comprehension in noise. The hearing aid is caught in a bind: the harder it works on audibility, the more it risks the very clarity it is trying to deliver. When that trade-off can no longer be won, the implant — which removes the distorting cochlea from the path — becomes the better instrument.

Wide dynamic range compression is the tool that fits speech into the narrowed window — and it genuinely raises the audibility of soft sounds (green). But it has a price: the harder it squeezes, the more it flattens the intensity contrasts between consonants and vowels that carry meaning, and it can worsen speech in noise (red). At high ratios the two curves cross — you have bought audibility at the cost of clarity. This is the chapter's “fix becoming the problem,” shown as a dial. Schematic.

Case 9.9 · When louder adds nothing

A patient's aided speech score has plateaued: making speech fully audible across the spectrum no longer improves understanding, and increasing compression seems to make noise worse.

How is this best understood?

Self-assessment — Module 92 questions

Question 1 · Trainee

Why do audibility-based models (the Speech Intelligibility Index) break down in severe loss?

Question 2 · Clinician

How can wide dynamic range compression, used to maximise audibility, backfire?

Tracked locally in your browser — see /progress for the dashboard.