Voice Disorders: Classification, Assessment, and Acoustic Characterization

Most clinical texts group voice disorders by underlying mechanism. The four-way grouping below is common, but it is a convention rather than a settled taxonomy: sources differ on where particular conditions belong, and the categories overlap in practice. Vocal hyperfunction, for instance, is behavioral in origin yet can produce structural lesions, which then provoke further compensatory behavior.

Voice disorders are common, though prevalence estimates vary widely with how the question is asked. In a random telephone survey of 1,326 adults in Iowa and Utah, Roy and colleagues (2005) found a point prevalence of 6.6% and a lifetime prevalence of 29.9%. National survey data show a higher and apparently rising figure: the US National Health Interview Survey put the proportion of adults reporting a voice problem in the preceding twelve months at 7.6% in 2012, and analyses of the 2022 survey placed it around 12% — roughly 30 million adults. The investigators note the increase may be connected to COVID-19 but treat that link as unconfirmed and in need of further study.

Estimates from other populations also run high — a Swedish cohort reported 16.9% self-assessed voice problems — and the spread across all of these figures reflects differences in case definition as much as any real difference between populations. The two 2022 analyses themselves reported slightly different rates, 11.7% and 12.2%, from the same survey, precisely because they defined a case differently.

Risk is not evenly distributed. In the Roy data, women, adults aged 40 to 59, and people with high vocal demands had elevated odds of chronic voice problems. Occupational voice users are the clearest case: a companion study of 1,243 teachers and 1,288 non-teachers found current voice problems in 11.0% of teachers against 6.2% of non-teachers, and lifetime prevalence of 57.7% against 28.8%. Prevalence also rises with age: a 2023 systematic review and meta-analysis estimated a pooled prevalence of 18.8% in older adults, rising to 33.0% among those living in institutional settings.

Two features of these data are worth holding onto. First, only a minority of affected people seek professional care, so clinic populations are not representative of the condition in the general population. Second, the spread across studies is a reminder that the boundary between an ordinary voice complaint and a voice disorder is drawn by definition, not discovered by measurement.

Contemporary voice assessment is multidimensional. No single measurement — perceptual, acoustic, aerodynamic, or visual — captures a voice disorder on its own, and the recommendation to combine them is one of the few points of broad agreement across professional guidance. What follows outlines how a voice evaluation is organized, then takes each dimension in turn and links to the guides that treat it in detail.

Patients arrive by two main routes: referred to a laryngologist for a voice complaint, or referred to a speech-language pathologist for assessment and therapy. The two disciplines answer different questions, and a complete evaluation usually requires both. The laryngologist visualizes the larynx and establishes the diagnosis. The SLP characterizes how the voice functions, what behaviors sustain the problem, and what it costs the patient — and, where behavioral treatment is indicated, delivers it.

The SLP evaluation itself follows a fairly stable sequence. It opens with case history and a patient-reported outcome measure, moves to auditory-perceptual rating, then to instrumental measurement, and closes with integration of the findings into a plan and a documented baseline.

The recorded stimulus set almost always includes both a sustained vowel and connected speech, because they answer different questions. A sustained vowel isolates phonation from articulation and is what most published acoustic norms are derived from; connected speech — a standard reading passage or the CAPE-V sentences — reflects how the voice behaves in use, including onsets, pitch variation, and the effects of running out of breath. A voice can look unremarkable on a sustained vowel and be plainly disordered in conversation.

Integration is the part that matters most and is hardest to systematize. Findings sometimes converge — a strained perceptual quality, elevated effort on history, and acoustic evidence of an atypical signal all pointing the same way. Just as often they diverge: a patient with a severe handicap score and near-normal acoustic values, or the reverse. Divergence is a finding, not an error. It usually indicates that the dimensions are measuring different things, which is precisely why the battery is multidimensional.

Whatever the protocol, its value depends on repeating it identically. A baseline is only useful if the follow-up is collected with the same equipment, the same distance, the same tasks, and the same analysis settings.

Assessment begins with onset, course, vocal demands, medical and surgical history, reflux and respiratory symptoms, and the patient's own account of the impact. Validated self-report instruments — the Voice Handicap Index and the Voice-Related Quality of Life measure among them — quantify that impact. They are the only part of the battery that measures what the disorder costs the person, and they correlate only loosely with acoustic findings. That dissociation is informative rather than a defect.

Trained listener judgment remains the reference against which acoustic measures are validated. CAPE-V and GRBAS are the two dominant protocols. Their reliability depends on training, anchors, and stimulus control, and no acoustic parameter maps one-to-one onto a perceptual dimension — a point developed at length in the companion guide.

Acoustic analysis quantifies properties of the recorded signal. Its value is that it is numerical, repeatable, and independent of listener drift, which makes it well suited to documenting change across a course of therapy or around a surgical intervention. The practical question is not which measure is best in general, but which measure answers the question being asked.

Acoustic measures are properties of a recording, not of a larynx. Microphone type and distance, ambient noise, sampling rate, and any processing applied in transit all influence the resulting numbers, and several widely used parameters are sensitive enough to these factors that a protocol change can outweigh a genuine clinical change. Two guides cover the cases that arise most often in practice.

Aerodynamic measures — subglottal pressure estimates, mean phonatory airflow, laryngeal airway resistance — describe the driving conditions of phonation rather than the resulting sound, and require instrumentation beyond a microphone. Laryngeal imaging, principally videostroboscopy and increasingly high-speed videoendoscopy, visualizes the folds and their vibratory behavior directly.

Imaging is the only part of the battery that identifies the lesion or the neurological deficit, and it falls to the laryngologist or to a clinician credentialed for endoscopy. Neither of these dimensions is within the scope of acoustic analysis, and neither can be substituted by it.

Management follows the diagnosis rather than the symptom, which is why the diagnostic step cannot be skipped. Three broad routes exist, frequently combined.

Behavioral voice therapy, delivered by a speech-language pathologist, is first-line for functional disorders and for phonotraumatic lesions, and has an established evidence base across several approaches, including resonant voice therapy, vocal function exercises, and intensive programs for Parkinson's disease. It also has a role before and after phonosurgery.

Medical management addresses contributing conditions — reflux, allergy, inflammatory disease, endocrine factors — or treats the disorder directly, as with botulinum toxin injection for spasmodic dysphonia.

Surgical management covers phonomicrosurgery for benign lesions, injection and framework laryngoplasty for glottal insufficiency, and oncological treatment where malignancy is present.

Selecting among these is a clinical decision for the treating team. This guide does not attempt to inform it.

The defensible claim for acoustic analysis in voice disorders is narrow and worth stating precisely. Acoustic measures quantify phonatory characteristics of a recorded voice sample. Repeated under a stable protocol, they document change over time — before and after therapy, around a surgical intervention, across the course of a progressive neurological condition. That is a real contribution, and one that perceptual judgment alone provides less reliably.

What acoustic measures do not do is identify a cause. A low CPPS value is consistent with nodules, with paralysis, with muscle tension dysphonia, and with a great many other things. Published cutoffs describe the distributions observed in particular validation samples, recorded with particular equipment and protocols; they are reference points for interpretation by a clinician, not verdicts about an individual. Applied to a patient whose sample was collected differently from the validation cohort, they may not transfer at all.

PhonaLab computes these measures in the browser using established Praat algorithms, with the provenance of each cutoff and each method stated on the results themselves. Audio is processed in memory and never stored. The platform reports measurements; the clinical interpretation stays with the clinician.