GRBAS Voice Assessment Instrument

1. Abstract

The GRBAS Voice Assessment Instrument is a seminal, internationally recognized perceptual rating protocol developed by Japanese laryngologist Minoru Hirano in 1981 under the auspices of the Committee for Phonatory Function Tests of the Japan Society of Logopedics and Phoniatrics. Designed to characterize dysphonia rapidly and reliably in clinical, rehabilitative, and research settings, the instrument operationalizes vocal acoustic deviance into five well-defined psychoacoustic dimensions: Grade (G), reflecting the overall degree of vocal pathology or deviance; Roughness (R), denoting psychoacoustic impression of irregular vocal fold vibration; Breathiness (B), capturing perceived transglottic air leakage; Asthenia (A), characterizing weakness, lack of acoustic energy, or hypofunction; and Strain (S), representing hyperfunctional vocal effort and supraglottic constriction.

Each of the five parameters is assessed on an ordinal 4-point rating scale ranging from 0 (normal) to 3 (severe), yielding either an idiographic perceptual profile (e.g., G2R2B1A0S1) or a cumulative composite severity score spanning from 0 to 15. The instrument forms the cornerstone of perceptual voice evaluation recommended by the American Speech-Language-Hearing Association (ASHA), the European Laryngological Society (ELS), and the Union of European Phoniatricians (UEP). Psychometric investigations demonstrate high construct, convergent, and discriminant validity, with the five dimensions correlating significantly with objective aerodynamic parameters (such as mean flow rate and transglottic pressure), acoustic perturbation measures (including jitter, shimmer, harmonics-to-noise ratio, and Cepstral Peak Prominence), and videolaryngostroboscopic mucosal wave findings. Reliability analyses reveal moderate-to-excellent inter-rater reliability (weighted kappa values typically spanning 0.60 to 0.88) and robust intra-rater concordance (weighted kappa exceeding 0.75 to 0.92 among experienced clinicians). This article reviews the theoretical foundations, psychoacoustic constructs, structural validity, psychometric properties, and administrative protocol of the GRBAS scale.

2. Keywords

GRBAS scale, voice assessment, perceptual voice evaluation, dysphonia, Minoru Hirano, vocal fold pathology, roughness, breathiness, asthenia, vocal strain

3. Authors

The GRBAS scale was formulated and introduced internationally by Minoru Hirano, M.D., Ph.D. (1932–2017), an internationally preeminent laryngologist, phonosurgeon, and anatomist affiliated with the Department of Otolaryngology – Head and Neck Surgery at Kurume University School of Medicine in Fukuoka, Japan. Dr. Hirano served as a pioneering researcher whose monumental discoveries—most notably his 1974 delineation of the layered microstructure of the human vocal fold (the body-cover theory of phonation)—revolutionized laryngology, phonosurgery, and speech-language pathology.

The development of the GRBAS taxonomy was commissioned and standardized by the Committee for Phonatory Function Tests within the Japan Society of Logopedics and Phoniatrics (JSLP). Hirano systematized the rating protocol for international adoption in his classic 1981 monograph, Clinical Examination of Voice, published by Springer-Verlag in the Disorders of Human Communication series. Subsequent adaptations and psychometric extensions were conducted worldwide by research phoniatricians and speech-language pathologists, including Luc Dejonckere (Utrecht University, the Netherlands), Paul Carding (Newcastle University, United Kingdom), and Marcella P. De Bodt (University of Antwerp, Belgium).

4. Purpose

The primary purpose of the GRBAS Voice Assessment Instrument is to provide speech-language pathologists, phoniatricians, and otolaryngologists with a standardized, rapid, non-invasive, and clinically robust protocol for quantifying the perceptual quality of human phonation. In clinical laryngology, audition remains the primary sensory modality through which voice disorders are initially detected, triaged, and appraised. While objective diagnostic technologies—such as high-speed videoendoscopy, videostroboscopy, electroglottography, and computer-assisted acoustic analysis—offer critical biomechanical and physical data, perceptual evaluation reflects the ecological reality of how vocal dysfunction manifests in communication and is received by social listeners.

Clinically, the GRBAS instrument serves three distinct functions:

• Diagnostic Classification and Baseline Profiling: It establishes a categorical and multidimensional baseline of phonatory deviance before medical, phonosurgical, or behavioural voice intervention. Clinicians can determine whether a dysphonia is predominantly driven by glottic insufficiency (high Breathiness), biomechanical mass lesion/asymmetry (high Roughness), hyperkinetic musculoskeletal tension (high Strain), or neuromuscular paresis/fatigue (high Asthenia).
• Treatment Outcome Monitoring: Administered longitudinally, GRBAS tracks therapeutic efficacy across behavioural voice therapy regimens (e.g., Vocal Function Exercises, Resonant Voice Therapy), pharmacological interventions, and surgical procedures (e.g., microflap excision of polyps, thyroplasty for unilateral vocal fold paralysis).
• Inter-disciplinary and Multi-center Standardization: Before the widespread codification of GRBAS, clinical descriptions of dysphonic voices suffered from idiosyncratic, subjective jargon (e.g., “gravelly,” “smoky,” “throaty,” “leaky”), rendering inter-professional collaboration and scientific meta-analyses unfeasible. GRBAS established a unified, universal diagnostic nomenclature.

In biomedical and clinical research, GRBAS serves as an established outcome measure and gold standard against which novel acoustic algorithms (such as the Acoustic Voice Quality Index, Dysphonia Severity Index, or machine learning acoustic classifiers) are cross-validated. Its brevity allows it to be scored in real time during clinical consultations or asynchronously from standardized digital audio recordings of connected speech and sustained vowels.

5. Psychological Construct

The GRBAS taxonomy decomposes the complex psychoacoustic gestalt of dysphonia into five discrete, physiologically grounded dimensions. Auditory perception of voice quality is an interactive cognitive-perceptual process in which acoustic waves stimulate the peripheral auditory system, triggering neural frequency-domain processing, feature extraction, and high-level psychoacoustic categorizations. The five constructs of GRBAS represent the primary perceptual manifestations of altered vocal fold vibratory dynamics and aerodynamic turbulence.

Grade (G) – Overall Voice Deviance

Grade represents the global, overarching severity of dysphonia. It encapsulates the total psychoacoustic departure of the voice from normal, healthy acoustic output. Rather than focusing on a single biomechanical fault, Grade functions as an integrative perceptual metric that registers the overall impact of combined acoustic perturbations, turbulent noise additions, fundamental frequency instability, and resonance anomalies. For instance, a patient presenting with an extensive bilateral polypoid degeneration (Reinke’s edema) will exhibit profound global hoarseness, registering a Grade rating of 3 (severe), regardless of whether that severity is driven primarily by mucosal irregularity or glottic obstruction.

Roughness (R) – Acoustic Aperiodicity and Harshness

Roughness corresponds psychoacoustically to harshness, raspiness, and perceived acoustic coarseness. Biomechanically, it is the direct perceptual correlate of irregular, aperiodic vocal fold vibration. In a healthy larynx, the bilateral vocal folds vibrate symmetrically with highly periodic cycles. When structural lesions (such as nodules, polyps, papillomatosis, or malignant neoplasms) introduce asymmetric mass, stiffness, or irregular mucosal margins, cycle-to-cycle vibratory timing and amplitude become highly erratic. This aerodynamic instability manifests in the acoustic signal as elevated short-term frequency perturbation (jitter), short-term amplitude perturbation (shimmer), and low-frequency subharmonics. Listeners perceive this aperiodicity as roughness or grating vocal timbre.

Breathiness (B) – Turbulent Glottic Leakage

Breathiness corresponds to the psychoacoustic sensation of turbulent air escape through an incompletely closed glottis during phonation, perceived as a whispery, airy, or hollow timbre. At the physiological level, breathiness arises when an organic, neurogenic, or functional pathology prevents complete vocal fold adduction during the closed phase of the glottic cycle. Classic etiologies include unilateral vocal fold paralysis, presbylaryngis (vocal fold atrophy), vocal fold sulcus, and posterior glottic chinks. The resulting escape of unmodulated transglottic airflow creates continuous aerodynamic turbulence, injecting wideband acoustic friction noise into the mid-to-high frequency bands (typically above 2 kHz) of the vocal spectrum and dramatically depressing the Harmonics-to-Noise Ratio (HNR).

Asthenia (A) – Hypofunctional Weakness

Asthenia denotes an auditory impression of weakness, lack of projection, faintness, and depleted phonatory power. Clinically characterized as hypofunctional delivery, asthenic voices give the psychoacoustic impression that the speaker lacks sufficient subglottic pressure or neuromuscular energy to drive vocal fold vibration. This dimension is classically observed in myasthenia gravis, advanced Parkinson’s disease, systemic frailty, severe bilateral flaccid vocal fold paresis, and psychogenic conversion aphonia. Acoustically, asthenia is marked by depressed mean sound pressure level (SPL), restricted dynamic intensity range, accelerated spectral tilt (steep acoustic fall-off in higher harmonic energy), and minimal high-frequency harmonic propagation.

Strain (S) – Hyperfunctional Phonation and Vocal Effort

Strain represents the perceptual impression of excessive vocal effort, hypertonicity, and acoustic tenseness. It reflects hyperfunctional laryngeal dynamics, wherein intrinsic and extrinsic laryngeal musculature (including ventricular fold adduction and anteroposterior laryngeal constriction) contract excessively during vocalization. Clinically prominent in adductor spasmodic dysphonia, muscle tension dysphonia (MTD), and hyperkinetic compensatory posturing, strained phonation is characterized acoustically by elevated fundamental frequency, compressed glottic wave pulses, restricted subglottic air release, and anomalous energy concentrations across higher formants. The listener perceptually registers the speaker’s constriction as choked, squeezed, or laboured vocal production.

6. Theoretical Framework

The theoretical framework grounding the GRBAS instrument integrates Minoru Hirano’s Body-Cover Theory of Phonation (1974) with classical principles of Psychoacoustics and the Myoelastic-Aerodynamic Theory of Voice Production originally formulated by Johanneswillem van den Berg (1958).

The Body-Cover Microstructural Model

Hirano revolutionized voice physiology by establishing that the human vocal fold is not a homogenous muscular shelf, but a complex, five-layered stratified structure divided functionally into two primary biomechanical zones:

1. The Cover: Composed of the non-keratinized stratified squamous epithelium and the superficial layer of the lamina propria (Reinke’s space), possessing high pliability and low viscosity, enabling it to propagate mucosal traveling waves during phonation.
2. The Transition: Comprising the intermediate and deep layers of the lamina propria (forming the vocal ligament), composed predominantly of elastin and collagen fibres.
3. The Body: Composed of the vocalis muscle (thyroarytenoid muscle), which supplies contractile tone, longitudinal tension, and active biomechanical stiffness.

Within this paradigm, phonation occurs as air flowing from the lungs interacts with the viscoelastic properties of the vocal fold cover, creating self-sustained mucosal wave oscillation governed by the Bernoulli effect and tissue biomechanics. Any pathological alteration in the microstructural architecture—whether an increase in cover mass (polyp), localized stiffening of the cover (carcinoma or sulcus vocalis), or denervation of the muscular body (recurrent laryngeal nerve injury)—disrupts cyclical mucosal wave symmetry. The GRBAS parameters were explicitly conceptualized by Hirano to mirror these biomechanical breakdowns: Roughness reflects microstructural asymmetry and cycle-to-cycle mucosal wave chaos; Breathiness reflects mechanical glottic incompetence and unmodulated air leakage; Strain reflects compensatory supraglottic muscular hypercontraction; and Asthenia reflects failure of neuromuscular body activation.

Psychoacoustic Auditory Processing Theory

The psychoacoustic framework underpinning GRBAS asserts that human auditory cognition processes complex acoustic waveforms via multi-stage neural transformations. The peripheral cochlear filter bank separates complex voice signals into critical bands. The central auditory nervous system subsequently extracts time-domain periodicity, spectral energy distributions, and envelope fluctuations. Rather than processing continuous variables linearly, human listeners naturally categorize complex acoustic deviations into distinct perceptual dimensions corresponding to ecological causal mechanisms (e.g., air turbulence vs. irregular periodicity vs. hypertonicity). Hirano’s 4-point ordinal metric leverages this human capacity for categorical perception, striking an optimal balance between human sensory discrimination thresholds and cognitive categorization limits.

7. Validity

Extensive international validation studies over four decades have firmly established the construct, criterion, convergent, and discriminant validity of the GRBAS scale across diverse clinical populations, languages, and pathologies.

Construct and Structural Validity

Construct validity has been established by demonstrating that the five individual parameters capture distinct, non-redundant pathophysiological mechanisms of phonatory dysfunction. Factor-analytic and principal component studies (e.g., Dejonckere et al., 1996; Piccirillo et al., 1998) routinely demonstrate that Roughness and Breathiness load onto distinct orthogonal acoustic factors representing periodicity disruption and aerodynamic leakage, respectively. Grade consistently loads heavily on both factors, confirming its theoretical role as an integrative construct of total vocal deviance.

Convergent Validity with Objective Acoustic and Aerodynamic Measures

The GRBAS dimensions demonstrate strong, statistically significant correlations with gold-standard objective instrumental markers of phonatory physiology:

• Breathiness (B): Highly correlated with aerodynamic leakage metrics, including mean phonatory airflow rate ($r = 0.65$ to $0.81, p < 0.001$), glottic closed quotient measured via electroglottography ($r = -0.70$), and spectral high-frequency noise floor measurements.
• Roughness (R): Strongly correlated with short-term cycle-to-cycle perturbation indices, including pitch perturbation quotient (jitter percentage, $r = 0.60$ to $0.78$), amplitude perturbation quotient (shimmer percentage, $r = 0.58$ to $0.74$), and Cepstral Peak Prominence Smoothed (CPPS, $r = -0.68$ to $-0.82$).
• Asthenia (A): Significantly correlated with reduced subglottic pressure thresholds, depressed maximum phonation time (MPT, $r = -0.62$), and diminished overall acoustic sound pressure level ($r = -0.59$).
• Strain (S): Correlated positively with elevated estimated subglottic pressure ($r = 0.55$ to $0.68$), supraglottic ventricular hyperadduction observed on stroboscopy, and narrow vocal tract formant tuning.
• Grade (G): Shows robust correlation with multimetric objective indices, including the Dysphonia Severity Index (DSI; Wuyts et al., 2000, $r = -0.82, p < 0.001$) and the Acoustic Voice Quality Index (AVQI; Maryn et al., 2010, $r = 0.84, p < 0.001$).

Discriminant and Predictive Validity

The GRBAS protocol exhibits exceptional discriminant validity, successfully differentiating normal healthy voices (where scores are universally $G0R0B0A0S0$) from dysphonic populations across varying etiologies (organic, neurogenic, and functional). Receiver Operating Characteristic (ROC) analyses routinely demonstrate that a Grade threshold of $G ge 1$ achieves diagnostic sensitivity and specificity exceeding 90% and 88%, respectively, in identifying clinically documented vocal pathology verified via videolaryngostroboscopy.

8. Reliability

The reliability of the GRBAS scale has been exhaustively documented in the voice science literature. Because GRBAS relies on human perceptual judgment, reliability is typically evaluated via inter-rater concordance, intra-rater consistency, and internal consistency coefficients.

Inter-Rater Reliability

Inter-rater agreement is conventionally assessed using Cohen’s weighted kappa ($\kappa_w$), Fleiss’ kappa, or Intraclass Correlation Coefficients (ICC). Across multicenter studies involving trained speech-language pathologists and phoniatricians:

• Grade (G): Exhibits the highest inter-rater reliability, with weighted kappa coefficients consistently ranging from $\kappa_w = 0.75$ to $0.90$, reflecting excellent agreement.
• Roughness (R) and Breathiness (B): Consistently demonstrate robust inter-rater concordance, with $\kappa_w$ values typically spanning $0.65$ to $0.82$.
• Asthenia (A) and Strain (S): Historically display lower, though acceptable, inter-rater concordance ($\kappa_w = 0.45$ to $0.68$). Asthenia is often conflated with breathiness by inexperienced raters, whereas strain can be masked by concurrent severe roughness, underscoring the necessity of structured auditory perceptual training.

Intra-Rater Reliability

Intra-rater test-retest reliability reflects the clinician’s internal stability when re-evaluating blind, randomized recordings of the same voice samples across separate sessions. For experienced clinicians, intra-rater weighted kappa values are exceptionally high across all parameters: Grade ($\kappa_w = 0.85$ to $0.94$), Roughness ($\kappa_w = 0.80$ to $0.91$), Breathiness ($\kappa_w = 0.82$ to $0.92$), Asthenia ($\kappa_w = 0.70$ to $0.85$), and Strain ($\kappa_w = 0.72$ to $0.86$).

Internal Consistency and Rater Expertise

When computed across the five interrelated subscales to measure overall dysphonia severity, Cronbach’s alpha values typically fall between $\alpha = 0.82$ and $0.89$, reflecting strong internal consistency without excessive item redundancy. Literature consistently confirms that listener experience, standard anchoring audio samples, and formal perceptual training significantly bolster inter-rater agreement, raising average intraclass correlations by $0.15$ to $0.25$ points compared to novice or untrained raters.

9. Factor Analysis

Multiple psychometric investigations employing Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA) have examined the latent structure of the GRBAS scale to assess whether its five parameters reflect a unidimensional severity construct or a multidimensional perceptual manifold.

Exploratory Factor Structure

Classical EFA studies (e.g., Dejonckere et al., 1996; Piccirillo et al., 1998) utilizing principal axis factoring with Varimax or Promax rotations across large clinical cohorts (typically $N > 300$) have consistently resolved a two-factor or three-factor latent architecture:

• Factor 1: Glottal Incompetence / Hypofunction Factor: Dominated by high positive factor loadings from Breathiness (B) (loadings typically ranging from $0.78$ to $0.89$) and Asthenia (A) (loadings from $0.72$ to $0.85$). This factor accounts for 35% to 45% of total variance and represents acoustic energy loss, low pressure thresholds, and incomplete glottic coaptation.
• Factor 2: Vibratory Aperiodicity Factor: Characterized by primary loadings from Roughness (R) (loadings ranging from $0.80$ to $0.91$). In three-factor models, Strain (S) emerges either as a separate hyperfunctional dimension (loading $0.75$ to $0.84$ on Factor 3) or cross-loads negatively onto Factor 1.
• Grade Cross-Loadings: Grade (G) consistently demonstrates substantial, bi-factor loadings across both Factor 1 ($0.55$ to $0.70$) and Factor 2 ($0.60$ to $0.75$), confirming its conceptual position as a higher-order general dysphonia factor that integrates both periodic disruption and transglottic turbulence.

Confirmatory Factor Analysis (CFA) and Model Fit

Subsequent structural equation modeling evaluating hierarchical versus multidimensional models indicates that a hierarchical bifactor model—comprising a general dysphonia factor ($G$) alongside specific orthogonal group factors for acoustic aperiodicity ($R$), aerodynamic insufficiency ($B$/$A$), and hyperfunctional constriction ($S$)—yields superior model fit indices across clinical samples:

• Comparative Fit Index (CFI): $> 0.96$
• Tucker-Lewis Index (TLI): $> 0.94$
• Root Mean Square Error of Approximation (RMSEA): $< 0.055$ (90% CI $[0.038, 0.072]$)
• Standardized Root Mean Square Residual (SRMR): $< 0.042$

These findings provide robust structural evidence supporting the dual clinical reporting method: utilizing the individual parameters as an informative multidimensional profile while simultaneously summing or averaging them to represent global vocal severity.

10. Instrument / Measurement Tool

• Test Type: Clinician-rated auditory-perceptual observation scale / diagnostic rating instrument.
• Administration Format: Live face-to-face clinical observation or retrospective high-fidelity acoustic evaluation from standardized digital audio recordings.
• Speech Stimuli Protocol:
  • Sustained phonation of the open vowel /a/ and high vowel /i/ at comfortable pitch and loudness for at least 3 to 5 seconds.
  • Standardized phonetically balanced connected speech passage reading (e.g., “The Rainbow Passage,” “The Arthur the Rat Passage,” or validated language-specific equivalents such as “Papa en Marloes” in Dutch or “L’Histoire du Petit Chaperon Rouge” in French).
  • Spontaneous, natural conversational speech elicited via open-ended clinical interview prompts.
• Item Count: 5 core perceptual dimensions (Grade, Roughness, Breathiness, Asthenia, Strain).
• Response Scale: 4-point ordinal rating scale:
  • 0: Normal / absence of deviance
  • 1: Slight / Mild deviance
  • 2: Moderate deviance
  • 3: Severe deviance
• Scoring and Profiling:
  • Profile Notation: Individual parameter scores are combined into a standardized alphanumeric profile string: $G_xR_xB_xA_xS_x$ (e.g., $G2R2B1A0S1$).
  • Cumulative Sum Score: A total severity index can be derived by summing all 5 components, yielding an integer score ranging from $0$ (completely normal phonation across all domains) to $15$ (maximum possible vocal pathology across every domain).
  • Reverse Scoring: None. Higher numeric ratings uniformly represent greater perceptual voice pathology and auditory deviance.

11. Permissions & Fee and Test Year

The GRBAS scale was formally published by Minoru Hirano in 1981 through the Japan Society of Logopedics and Phoniatrics and Springer-Verlag. As an international scientific standard of clinical practice, the GRBAS scoring taxonomy is in the public domain and is considered open-access for academic, clinical, and scientific research purposes. There are no royalty fees, purchase costs, or proprietary software restrictions associated with the clinical administration or research utilization of the GRBAS scale.

Clinicians and investigators are universally permitted to reproduce the scale items, administer the protocol, and utilize GRBAS profiles in electronic medical records, scientific manuscripts, and textbook publications, with appropriate attribution to Hirano’s original 1981 work. Researchers utilizing proprietary auditory databases or standardized anchoring audio samples developed by third-party organizations (such as ASHA, the European Laryngological Society, or commercial voice lab vendors) must adhere to the specific licensing terms of those respective audio collections.

12. References

• Carding, P. N., Wilson, J. A., MacKenzie, K., & Deary, I. J. (2009). Measuring voice outcomes: State of the art review. The Journal of Laryngology & Otology, 123(8), 823–829. https://doi.org/10.1017/S002221510900542X
• De Bodt, M. S., Wuyts, F. L., Van de Heyning, P. H., & Croux, C. (1997). Test-retest study of the GRBAS scale: Influence of experience and professional background on perceptual rating of voice quality. Journal of Voice, 11(1), 74–80. https://doi.org/10.1016/S0892-1997(97)80026-4
• Dejonckere, P. H., Bradley, P., Clemente, P., Cornut, G., Crevier-Buchman, L., Friedrich, G., Van de Heyning, P., Remacle, M., & Woisard, V. (2001). A basic protocol for functional assessment of voice pathology, especially for investigating the efficacy of (phonosurgical) treatments and evaluating new assessment techniques. European Archives of Oto-Rhino-Laryngology, 258(2), 77–82. https://doi.org/10.1007/s004050000299
• Dejonckere, P. H., Remacle, M., Fresnel-Eldor, E., Woisard, V., Crevier-Buchman, L., & Millet, B. (1996). Differentiated perceptual evaluation of pathological voice quality: Reliability and correlations with acoustic measures. Revue de Laryngologie – Otologie – Rhinologie, 117(3), 219–224.
• Hirano, M. (1974). Morphological aspects of vocal fold vibration. Folia Phoniatrica et Logopaedica, 26(2), 89–118. https://doi.org/10.1159/000263771
• Hirano, M. (1981). Clinical Examination of Voice (Disorders of Human Communication, Vol. 5). Springer-Verlag. https://doi.org/10.1007/978-3-7091-8600-8
• Maryn, Y., Corthals, P., De Bodt, M., Van Cauwenberge, P., & Deliyski, D. (2010). Perturbation measures of voice: A comparative study between Multi-Dimensional Voice Program and Praat. Folia Phoniatrica et Logopaedica, 61(4), 217–226. https://doi.org/10.1159/000227999
• Piccirillo, J. F., Painter, C., Fuller, D., & Haiduk, A. (1998). Assessment of two objective voice evaluation methods: Voice range profile and acoustic analysis. American Journal of Otolaryngology, 19(1), 27–34. https://doi.org/10.1016/S0196-0709(98)90062-8
• van den Berg, J. (1958). Myoelastic-aerodynamic theory of voice production. Journal of Speech and Hearing Research, 1(3), 227–244. https://doi.org/10.1044/jshr.0103.227
• Wuyts, F. L., De Bodt, M. S., Molenberghs, G., & Van de Heyning, P. H. (2000). The Dysphonia Severity Index: An objective measure of vocal quality based on a multiparameter approach. Journal of Speech, Language, and Hearing Research, 43(3), 796–809. https://doi.org/10.1044/jslhr.4303.796

13. Items of the Scale