How Voice Biomarkers Reveal Mental Health
Understanding jitter, shimmer, and harmonic-to-noise ratio in acoustic analysis
What Are Voice Biomarkers?
Voice biomarkers are measurable acoustic features extracted from a person's speech that can indicate changes in their physical or psychological state. Just as blood pressure or heart rate serve as biomarkers for cardiovascular health, certain properties of the human voice serve as reliable indicators for mental and emotional conditions.
The human voice is produced by a complex interplay of respiratory, laryngeal, and articulatory systems. When a person experiences stress, anxiety, depression, or other conditions, these systems are affected in subtle but detectable ways. Muscle tension in the vocal folds changes, breathing patterns shift, and the neural control of speech production is altered.
Jitter: Micro-Variations in Pitch
Jitter measures the cycle-to-cycle variation in the fundamental frequency (pitch) of the voice. In a perfectly stable voice, each vibration of the vocal folds would be identical. In reality, there are always small fluctuations — and these fluctuations carry diagnostic information.
Elevated jitter levels are associated with increased muscle tension, neurological changes, and emotional distress. Research has shown that individuals experiencing anxiety or stress tend to exhibit higher jitter values compared to their baseline. Depression, conversely, can manifest as a different pattern of pitch instability.
Happo AI measures jitter at sub-millisecond precision, detecting variations too subtle for the human ear to perceive but significant enough to indicate changes in mental state.
Shimmer: Amplitude Instability
While jitter tracks pitch variation, shimmer measures the cycle-to-cycle variation in amplitude (loudness) of the voice. Shimmer reflects the regularity of vocal fold vibration and the efficiency of airflow through the larynx.
Conditions that affect respiratory control — such as panic, fatigue, or burnout — often produce elevated shimmer values. The connection between breathing patterns and emotional state is well-documented: shallow, irregular breathing during anxiety leads to less stable vocal fold vibration, which shimmer captures quantitatively.
Shimmer analysis is particularly valuable when combined with jitter measurements. The ratio and pattern of these two biomarkers together provide a more robust signal than either one alone, enabling the differentiation between conditions that might affect pitch and amplitude in distinct ways.
Harmonic-to-Noise Ratio (HNR)
The harmonic-to-noise ratio quantifies the proportion of periodic (harmonic) energy to aperiodic (noise) energy in the voice signal. A healthy, well-controlled voice produces a strong harmonic structure with minimal noise. When vocal fold vibration becomes irregular or incomplete, more turbulent airflow noise is introduced.
HNR is one of the most sensitive biomarkers for emotional and psychological state changes. Stress-induced muscle tension causes incomplete vocal fold closure, allowing air to escape and increasing the noise component. Depression can reduce the energy and control in speech production, similarly degrading the harmonic structure.
Happo AI uses HNR alongside jitter and shimmer to build a comprehensive vocal profile. By tracking these three biomarkers over the course of a conversation, the system can identify not just the presence of a condition, but how it fluctuates in real time — during moments of high pressure in a sales call or emotional breakthrough in a counselling session.
Beyond Individual Biomarkers
While jitter, shimmer, and HNR form the foundation of voice biomarker analysis, Happo AI extracts dozens of additional acoustic features from each audio segment. These include spectral characteristics, prosodic patterns (rhythm, tempo, intonation), and formant frequencies that together create a rich, multi-dimensional representation of the speaker's vocal state.
The true power of voice biomarker analysis lies in the combination and temporal tracking of these features. A single snapshot tells part of the story; tracking biomarker trajectories across an entire conversation reveals patterns that are clinically and commercially meaningful — from detecting escalating stress in a sales interaction to identifying gradual improvement in a patient's counselling journey.