Jitter and Shimmer
Jitter and shimmer represent the most widely used acoustic measures for quantifying short-term voice instability. These parameters capture cycle-to-cycle variations in fundamental period (jitter) and amplitude (shimmer), providing objective indices of voice quality that correlate with perceived roughness and vocal pathology. Despite their widespread clinical application, these measures require careful interpretation and understanding of their underlying assumptions, calculation methods, and limitations.
Fundamental Period Perturbation: Jitter
Jitter quantifies the variability in the duration of successive vocal fold vibratory cycles. In perfectly periodic phonation, each glottal cycle would have identical duration. Biological systems, however, exhibit inherent variability, with jitter measuring the magnitude of these period-to-period differences.
Physiological Basis
Several physiological mechanisms contribute to period perturbations. Neural control of laryngeal muscles involves motor unit recruitment governed by stochastic firing patterns, introducing cycle-to-cycle variability. Biomechanical asymmetries between left and right vocal folds create slight differences in tissue properties, mass distribution, and tension that affect oscillation timing. Aerodynamic turbulence in glottal airflow produces pressure fluctuations that perturb the driving forces for oscillation.
In healthy voices, these sources produce small random variations typically less than 1% of the mean period. Pathological conditions—vocal fold lesions, neurological impairment, inadequate glottal closure—amplify these variations through mechanisms including asymmetric oscillation, irregular mucosal wave propagation, and unstable aerodynamic coupling.
Jitter Calculation Methods
Multiple algorithms exist for calculating jitter, differing in how they normalize and express the period variations. The most common include:
Absolute Jitter measures the average absolute difference between consecutive periods:
Jitter (absolute) = (1/N-1) × Σ|Tᵢ - Tᵢ₊₁|
This produces a value in time units (microseconds or milliseconds). For a voice with 100 Hz average F₀ (10 ms period) and 50 μs absolute jitter, each period differs from its neighbor by 50 μs on average.
Jitter Percent normalizes by the mean period:
Jitter (%) = [(1/N-1) × Σ|Tᵢ - Tᵢ₊₁|] / T_mean × 100%
This dimensionless percentage enables comparison across different speakers and fundamental frequencies. The 50 μs absolute jitter example yields 0.5% jitter percent.
Relative Average Perturbation (RAP) compares each period to the average of itself and two neighbors:
RAP = (1/N-2) × Σ|Tᵢ - (Tᵢ₋₁ + Tᵢ + Tᵢ₊₁)/3| / T_mean
RAP provides greater resistance to transient irregularities by incorporating local averaging.
Five-Point Period Perturbation Quotient (PPQ) extends this smoothing to five consecutive periods:
PPQ = (1/N-4) × Σ|Tᵢ - (Tᵢ₋₂ + Tᵢ₋₁ + Tᵢ + Tᵢ₊₁ + Tᵢ₊₂)/5| / T_mean
The choice among these measures depends on the analysis goals. Simple jitter percent remains most common in clinical practice, while RAP and PPQ may provide better discrimination for specific pathologies or research applications.
Amplitude Perturbation: Shimmer
Shimmer quantifies cycle-to-cycle variations in vocal fold vibratory amplitude, typically assessed from the peak-to-peak amplitude of successive glottal pulses in the acoustic waveform. Like jitter, shimmer reflects the stability of the oscillatory system but focuses on amplitude rather than timing variations.
Physiological Sources
Amplitude perturbations arise from several mechanisms distinct from though related to period perturbations. Variations in subglottal pressure from respiratory muscle control and lung recoil properties create cycle-to-cycle differences in the driving force for oscillation. Inconsistent glottal closure—whether from biomechanical asymmetries, inadequate adduction, or tissue lesions—produces varying amounts of acoustic energy per cycle.
The relationship between glottal area waveform and acoustic output involves complex source-filter interactions. Even with constant subglottal pressure and glottal area variations, the resulting acoustic amplitude can vary due to time-varying vocal tract impedance and harmonic-formant proximity effects. These acoustic factors may contribute to measured shimmer independent of oscillatory instability.
Figure 11.6: Visualization of period perturbation (jitter) and amplitude perturbation (shimmer) in a speech waveform. The variations in cycle duration and peak amplitude are evident even in relatively normal phonation.
Shimmer Calculation Methods
Shimmer calculations parallel jitter methods, operating on amplitude rather than period measurements:
Shimmer (absolute) in decibels:
Shimmer (dB) = (1/N-1) × Σ|20 log₁₀(Aᵢ₊₁/Aᵢ)|
where Aᵢ represents the peak-to-peak amplitude of cycle i.
Shimmer Percent:
Shimmer (%) = [(1/N-1) × Σ|Aᵢ - Aᵢ₊₁|] / A_mean × 100%
Amplitude Perturbation Quotient (APQ) smooths across multiple cycles:
APQ = (1/N-10) × Σ|Aᵢ - (Aᵢ₋₄...Aᵢ₊₄)/11| / A_mean
where the denominator averages 11 consecutive amplitudes centered on cycle i.
The relationship between shimmer in percent and decibels is nonlinear. For small perturbations, approximate conversion follows:
Shimmer (dB) ≈ 0.115 × Shimmer (%)
Thus, 3% shimmer corresponds to roughly 0.35 dB cycle-to-cycle amplitude variation.
Normal Values and Clinical Thresholds
Establishing normal ranges for jitter and shimmer enables clinical interpretation. Numerous normative studies provide reference data, though specific values depend on measurement method, analysis software, vowel, pitch, and intensity.
Jitter Normal Values
Young healthy adults (age 20-40):
- Mean jitter: 0.3-0.6%
- Upper normal limit: 1.0%
- Mild pathology: 1.0-1.5%
- Moderate-severe pathology: >1.5%
Older adults (age 60+):
- Mean jitter: 0.5-1.0%
- Upper normal limit: 1.5%
- Pathology threshold: >2.0%
Age-related increases reflect tissue changes (collagen deposition, tissue stiffening), neural control degradation, and respiratory-laryngeal coordination decline.
Shimmer Normal Values
Young healthy adults:
- Mean shimmer: 1.0-2.0%
- Upper normal limit: 3.0%
- Mild pathology: 3.0-5.0%
- Moderate-severe pathology: >5.0%
Older adults:
- Mean shimmer: 2.0-3.5%
- Upper normal limit: 4.0%
- Pathology threshold: >6.0%
Shimmer generally shows greater absolute values and wider normal ranges than jitter, possibly reflecting greater susceptibility to respiratory and acoustic factors beyond oscillatory stability.
Clinical Significance
Elevated jitter and shimmer correlate with various pathological conditions, though the relationships remain imperfect and non-specific.
Structural Pathologies
Vocal fold lesions including nodules, polyps, cysts, and scarring typically elevate both jitter and shimmer. Asymmetric mass distribution disrupts left-right oscillation synchrony, while surface irregularities impair mucosal wave propagation. Incomplete glottal closure increases noise content and amplitude irregularity.
The magnitude of perturbation generally correlates with lesion size and impact on vibratory function, though considerable variability exists. Small lesions at critical locations (mid-membranous vocal fold) may produce greater perturbations than larger lesions elsewhere.
Neurological Pathologies
Parkinson’s disease, essential tremor, spasmodic dysphonia, and other neurological disorders affect voice through multiple mechanisms reflected in perturbation measures. Tremor superimposes low-frequency modulation often reflected in elevated jitter and shimmer, though distinguishing tremor from random perturbation requires spectral analysis.
Spasmodic dysphonia produces intermittent voice breaks and abrupt perturbations that may not be well-characterized by average jitter values computed across entire phonation samples. Task-dependent variation—comparing sustained vowels to connected speech—may provide better diagnostic discrimination.
Functional Voice Disorders
Muscle tension dysphonia and other functional disorders without structural lesions may show normal or paradoxically reduced perturbation measures despite clear perceived voice quality impairment. This dissociation highlights limitations of jitter and shimmer as sole indicators of voice quality, emphasizing the need for multidimensional assessment including perceptual evaluation.
Methodological Considerations and Limitations
Despite widespread use, jitter and shimmer measures face significant limitations requiring careful interpretation.
Period and Amplitude Detection Accuracy
All perturbation measures depend critically on accurate identification of fundamental period and amplitude for each cycle. This requires reliable pitch tracking and cycle marking, which becomes challenging when F₀ is low (<80 Hz), signals are noisy, or voice quality is severely impaired.
Detection errors—mistaking one period for two, missing a cycle, or inaccurate amplitude peak identification—introduce artifactual perturbation exceeding true biological variability. Commercial analysis systems employ various algorithms with differing robustness to noise and pathological voices, potentially producing incompatible results.
Signal Type Restrictions
Jitter and shimmer calculations assume Type 1 signals—nearly periodic waveforms with small perturbations around a stable mean. These measures become invalid or misleading for:
- Type 2 signals: Voice with strong modulation (tremor, heavy vibrato), subharmonics, diplophonia or other bifurcations
- Type 3 signals: Voice with no apparent periodic structure (chaotic or severely aperiodic phonation)
- Nonstationary segments of any type: pitch glides, onset and offset transients, voice breaks
Computing conventional perturbation measures on Type 2 or Type 3 signals produces numerical results, but these values lack meaningful interpretation. A voice with 6-Hz tremor will show elevated jitter, but this reflects intentional or pathological modulation rather than random cycle-to-cycle instability.
Microphone and Recording Factors
High-quality recordings prove essential for valid perturbation analysis. Microphone frequency response, ambient noise, analog-to-digital conversion parameters, and signal processing (filtering, compression) all influence measured values.
Standard protocols recommend:
- Sampling rate ≥20 kHz (preferably 44.1 kHz)
- 16-bit minimum resolution (24-bit preferred)
- High-pass filtering ≥50 Hz to remove low-frequency noise
- Low-pass filtering below Nyquist frequency
- Mouth-to-microphone distance 10 cm
- Quiet recording environment (ambient noise <50 dB SPL)
Failure to control these factors introduces measurement variability potentially exceeding true voice differences between speakers or conditions.
Phonation Task Effects
Perturbation values depend significantly on the phonation task. Sustained /a/ at comfortable pitch and loudness represents the clinical standard, but values differ substantially for:
- Different vowels (/i/, /u/ show different stability)
- Pitch extremes (low pitch increases perturbation)
- Loudness extremes (soft voice increases perturbation)
- Connected speech (shows greater variability than sustained vowels)
Comparing values across studies or clinical evaluations requires matching phonation tasks. Mixed-task protocols may increase diagnostic information but complicate interpretation.
Relationship Between Jitter and Shimmer
Jitter and shimmer show moderate positive correlation (r ≈ 0.4-0.6) in typical voices, suggesting partial independence of period and amplitude perturbation mechanisms. Some pathologies preferentially affect one measure: oscillatory asymmetries may increase jitter more than shimmer, while inadequate closure affects shimmer more than jitter.
This partial independence supports measuring both parameters rather than assuming one predicts the other. Combined interpretation—considering jitter-shimmer patterns rather than isolated values—may improve diagnostic discrimination, though more research is needed to establish pattern-pathology relationships.
Summary
Jitter and shimmer quantify cycle-to-cycle variations in fundamental period and amplitude, providing widely used indices of voice stability. Multiple calculation methods exist, with jitter percent and shimmer percent remaining most common. Normal values typically fall below 1% for jitter and 3% for shimmer in young adults, with age-related increases and pathological elevation beyond clinical thresholds.
These measures correlate imperfectly with vocal pathology, showing good sensitivity but limited specificity. Methodological limitations include dependence on accurate period detection, restriction to nearly periodic signals, sensitivity to recording conditions, and task-dependent variability. Jitter and shimmer provide valuable objective information when interpreted within the context of perceptual assessment, case history, and visual examination, but should not serve as sole diagnostic indicators.
Key Takeaways
- ✅ Jitter measures fundamental period variability; shimmer measures amplitude variability on a cycle-to-cycle basis
- ✅ Normal jitter values typically remain below 1%, shimmer below 3%, with higher thresholds for older adults
- ✅ Multiple calculation methods exist (absolute, percent, perturbation quotients), producing different numerical values
- ✅ Elevated perturbations correlate with vocal pathology but show imperfect sensitivity and specificity
- ✅ Measures require nearly periodic signals (Type 1); results become invalid for tremor, vibrato, or highly irregular voices
- ✅ Accurate measurements depend on high-quality recordings and precise period/amplitude detection algorithms
- ✅ Phonation task significantly affects values; sustained vowels at comfortable pitch/loudness provide standard conditions
- ✅ Jitter and shimmer show partial independence, supporting measurement of both parameters for comprehensive assessment
Related Topics
Further Reading
- Titze, I. R., Horii, Y., & Scherer, R. C. (1987). Some technical considerations in voice perturbation measurements. Journal of Speech and Hearing Research, 30(2), 252-260.
- Pinto, N. B., & Titze, I. R. (1990). Unification of perturbation measures in speech signals. Journal of the Acoustical Society of America, 87(3), 1278-1289.
- Deliyski, D. D., Shaw, H. S., & Evans, M. K. (2005). Adverse effects of environmental noise on acoustic voice quality measurements. Journal of Voice, 19(1), 15-28.
- Brockmann-Bauser, M., & Drinnan, M. J. (2011). Routine acoustic voice analysis: Time to think again? Current Opinion in Otolaryngology & Head and Neck Surgery, 19(3), 165-170.