Glossary

Voice and acoustics, in plain language.

The terms you'll see on the dashboard, defined the way we'd explain them to a curious friend. Search to jump.

A

AGC: Automatic gain control. Browser microphone processing that we deliberately disable because it distorts the time-domain signal pitch detection relies on.

Cents: 1/100 of a semitone. The unit we use for pitch deviation: ±25¢ is the Pitch Hold tolerance window, ±50¢ is the karaoke target.
CPPS: Cepstral peak prominence smoothed. The most modern voice-quality scalar — robust to background noise, correlated with breathiness and dysphonia severity.

Diadochokinesis (DDK): Rapid repetition of pa-ta-ka. Measures articulatory speed and is a standard speech-pathology screen.

F0: Fundamental frequency. The acoustic correlate of perceived pitch. Measured in Hz.
F1, F2, F3: First three formants. Resonant peaks of the vocal tract that determine vowel identity and tone colour. Estimated via LPC.
Filler: Um, uh, like, you know. Detected automatically in continuous speech in Volume Coach (beta).

GNE: Glottal-to-noise excitation ratio. Discriminates harmonic from noise components in the signal — sensitive to incomplete glottal closure.
GRBAS: Hirano 1981 perceptual scale: Grade, Roughness, Breathiness, Asthenia, Strain. Used as a self-rating in Voxavia.

HNR: Harmonics-to-noise ratio. Higher means a clearer harmonic structure; lower means more noise — often associated with breathiness or roughness.

Jitter: Cycle-to-cycle variation in pitch period. A classic voice-quality metric — elevated values can indicate vocal instability.
JND: Just-noticeable difference. The smallest pitch difference a listener can reliably detect. Voxavia estimates yours in cents via a 2AFC staircase.

LMRVT: Lessac-Madsen Resonant Voice Therapy. Five-phase protocol: hum, chant, vowels, words, phrases.
LPC: Linear predictive coding. The estimation method we use for formants every ~200 ms.
LSVT-LOUD: Lee Silverman Voice Treatment. Evidence-based protocol for hypophonia in Parkinson's disease — sustained loud phonation, glides, and functional phrases.
LTAS: Long-term average spectrum. The slope reflects vocal quality and projection over a longer window than a single take.

MDVP: Multi-Dimensional Voice Program. Long-standing clinical voice analysis software whose published norms we calibrate against.
MFCC: Mel-frequency cepstral coefficients. Widely used in speech analysis — Voxavia uses them for vowel-shape and accent work.
MPT: Maximum phonation time. The longest you can sustain a vowel on a comfortable pitch in one breath. A core breath/aerodynamic measure.

Passaggio: The pitch zone where a singer's voice transitions between registers. Voxavia maps yours by detecting break clusters across glides.
Praat: Open-source phonetics software widely used in voice research. Voxavia's DSP parameters are calibrated against Praat outputs.
Prosody: Pitch, loudness, and timing patterns that carry meaning beyond words. Reading expressiveness scores against pitch variance and dynamic range.

RSI: Reflux Symptom Index (Belafsky 2002). 9-item self-rating for laryngopharyngeal reflux.

Shimmer: Cycle-to-cycle variation in amplitude. The amplitude analogue of jitter.
Singer's formant: An energy peak around 2.8–3.4 kHz that lets unamplified voices project over an orchestra. Voxavia scores how present it is in your tone.
SNR: Signal-to-noise ratio. Voxavia uses an SNR gate so that your volume meter only counts when you're audibly above the room's noise floor.
Subharmonic ratio: Detects period-doubling — a sign of diplophonia or vocal-fold pathology.
S/Z ratio: Maximum sustained /s/ divided by maximum sustained /z/. A simple aerodynamic screen for glottal closure.

VHI: Voice Handicap Index. Patient-reported measure of how voice problems affect daily life — physical, functional, and emotional sub-scales.
Vibrato: Periodic modulation of pitch. Voxavia measures rate (Hz), extent (cents), and regularity.
Vowel space: A 2D plot of F1 vs F2. Each vowel sits in its own region — shape work targets clarity by hitting IPA-anchored targets.

YIN: Time-domain autocorrelation pitch detector. Voxavia's foundation — implemented via the Pitchy library.

Voxavia · 2026app.voxavia.com

Voxavia runs in your browser. There's nothing to install and no account needed to try it.

No account · No installDSP v1.0 · 2026