8Psychoacoustic Measurement (pitch, loudness, MML, RI)

FFrom bedside estimate to instrumented battery

Chapter 4 covered the quick bedside techniques. This module is the instrumented version: a standardised, audiometer- or software-controlled battery that yields numbers a second clinician could reproduce. The aim is to characterise what the patient hears — the percept — as distinct from how much it bothers them, which is the job of the questionnaires in the next module.

The core measures are pitch matching, loudness matching, the minimum masking level, residual inhibition and the tinnitus spectrum/likeness rating. Together they calibrate sound therapy (such as tinnitus retraining therapy), document stability or change over time, and provide the outcome variables for clinical trials [2014].

A crucial caveat sets the tone for the whole module: these measures describe the sound, not the suffering. A faint tinnitus matched at 5 dB sensation level can be devastating, and a loud one trivial. Psychoacoustics and patient-reported outcomes answer different questions and must be read together.

TPitch and loudness matching

Pitch matching presents pure tones (or narrow-band noise) and asks the patient to find the one that most resembles their tinnitus. The matched pitch usually lies in the high-frequency region, often around 4–8 kHz or higher, and tends to coincide with the edge of the patient’s hearing loss — evidence for the deafferentation/edge model [2008]. A common pitfall is octave confusion, where the patient matches a tone one octave away from their true percept; an octave-confusion check is part of any rigorous protocol.

Loudness matching raises a tone (usually at or near the matched pitch) until its loudness equals the tinnitus. Reported in dB sensation level (dB above that ear’s threshold at that frequency), most tinnitus matches at only 5–15 dB SL — strikingly faint given the distress it causes. Reporting in dB SL rather than dB HL is essential, because it corrects for the patient’s own threshold and is what makes the famous loudness-distress mismatch visible.

Both measures vary across sessions and patients, so single values are interpreted with caution and serial testing is preferred for tracking [2014].

CMinimum masking level and residual inhibition

The minimum masking level (MML) is the lowest level of a masking noise (white, pink or narrow-band) that just covers the tinnitus. It classifies tinnitus as easily masked, hard to mask, or unmaskable, and a low MML predicts a better response to sound-therapy masking, whereas a high or unmaskable result hints at more central generation [2014]. The MML and the loudness match together also feed the older masking-curve classifications.

Residual inhibition (RI) is the temporary suppression of tinnitus after a period of masking sound is switched off. Responses are graded — positive-complete (tinnitus gone), positive-partial (reduced), none, or paradoxical rebound (worse) — and the suppression typically lasts seconds to minutes. RI is informative because it demonstrates that the tinnitus signal is dynamic and modifiable, and the depth/duration of RI helps predict who may benefit from sound therapy [2008].

Importantly, RI functions overlap closely with the tinnitus spectrum and the region of hearing loss, reinforcing that the percept is generated by the same deafferented frequency channels — a finding with direct mechanistic and therapeutic meaning [2008].

CTinnitus spectrum, standardisation and automation

A single pitch match poorly captures a tinnitus that sounds like noise or a band of frequencies. The tinnitus spectrum / likeness rating asks the patient to rate, frequency by frequency, how closely each test tone resembles their tinnitus, producing a profile rather than a point estimate. This often reveals that the tinnitus spans the whole region of hearing loss, not a single tone [2008].

The chief limitation of psychoacoustic measurement has always been reliability: results drift between sessions and examiners. Standardisation efforts and computer-automated, forced-choice protocols substantially improve test–retest reliability for matching, maskability and RI, which is why automated systems are now favoured for research and increasingly for clinic use [2013].

The clinical takeaway: choose a standardised (ideally automated) protocol, report loudness in dB SL, run an octave-confusion check, and read every number as a description of the percept that must be combined with the patient’s self-reported impact before acting [2014].