Rate My Voice

What the Speaking Voice Test Measures

The single number at the top is a weighted combination of seven separate measurements. Each one is worth understanding on its own, because they fail for different reasons and improve through different kinds of practice. This page shows all seven, and the results panel gives you your own figure for each.

Clarity — 24% of the score

This is your harmonic-to-noise ratio: how much of your sound is clean, periodic vocal tone versus breath, turbulence and room noise. It carries the heaviest weight because it is what listeners consciously notice first. A healthy voice recorded in a quiet room usually measures above 15 dB; above 20 dB is broadcast territory. Under 10 dB something is wrong, and nine times out of ten that something is the room rather than the larynx.

Pitch stability — 20%

Measured as jitter: the percentage by which each vocal fold vibration cycle differs in length from the one before it. This is measured cycle by cycle in the time domain, not frame by frame, because averaging across a 40-millisecond frame smooths out exactly the variation being looked for. Healthy voices sit under 1%. Above 2% the roughness becomes audible as a creaky or unsteady quality.

Volume control — 14%

Shimmer, the amplitude equivalent of jitter, expressed in decibels of cycle-to-cycle variation. High shimmer sounds like a voice that wavers or lacks breath support. It is often the first thing to deteriorate when you are tired or speaking at the very bottom of your range.

Expressiveness — 14%

How far your pitch moves around its own median, measured in semitones rather than hertz — the ear hears pitch logarithmically, so a raw hertz spread would unfairly reward high voices. Around three semitones of movement reads as natural conversational speech. Near zero reads as monotone. Very large values in a speaking take usually mean instability rather than expression, so this sub-score peaks in the middle rather than at the top.

Tone brightness — 14%

The spectral centroid, or centre of gravity of your frequency spectrum. Low values sound dark and muffled; very high values sound thin or sibilant. The comfortable middle sits around 1900 Hz. A muffled reading is frequently the microphone position rather than the voice — speaking across a laptop lid, or into a phone lying face down on a desk, both drag this figure down sharply.

Delivery flow — 8%

The proportion of your take that was fully voiced rather than breathy, whispered or silent. Long pauses, trailing sentence endings and heavy breathiness all reduce it.

Range used — 6%

The span between your 5th and 95th percentile pitch in this take. It carries little weight when rating speech, because a normal sentence simply does not need a wide range. It is weighted far more heavily on the singing test.

How to Actually Improve Each Sub-Score

Two of these sub-scores respond to your recording setup within seconds, and the rest respond to practice over weeks. It is worth knowing which is which before you conclude anything about your voice.

Fixable in seconds: clarity and tone brightness. Move to a quieter room, sit 15 to 25 centimetres from the microphone, speak across it rather than directly into it, and turn off any headset noise cancelling. It is common to gain a full point on the total score from setup changes alone, which tells you something important — a low first result usually is not a verdict on your voice.

Responds to warm-up: pitch stability and volume control. Jitter and shimmer are both measurably worse on a cold voice first thing in the morning. Five minutes of gentle humming and lip trills, then re-record, and you will typically see both improve.

Responds to weeks of practice: expressiveness and delivery flow. These are speaking habits, not physiology. Reading aloud for ten minutes a day — deliberately varying pace and pitch, and finishing sentences with support rather than letting them trail into breath — moves both over a month. This is the area where recording yourself weekly and keeping the numbers is genuinely useful.

Largely fixed: your average pitch. Habitual speaking pitch can be shifted with training, but it is anatomically anchored and not something to chase for the sake of a score.

Why Recording Conditions Matter So Much

Because clarity carries the largest single weight, the acoustics of your room influence your result more than most people expect. A running fan, an open window onto a street, or a laptop fan spinning up under load each add broadband noise that sits directly in the frequency band being measured, and each can cost close to a full point.

The tool requests your microphone with echo cancellation, noise suppression and automatic gain control explicitly disabled, because all three actively destroy the micro-variations that jitter and shimmer are built from. Aggressive noise suppression in particular can make a voice score artificially high on stability while gutting its clarity. Some processing happens in the headset firmware or the operating system where the browser has no control, which is why a wired microphone or a built-in laptop mic usually gives a more faithful reading than a Bluetooth headset.

One practical consequence: do not compare your score against a friend's unless you both recorded on similar hardware in similar rooms. Compare against your own previous takes instead.

What This Score Is Not

It is an acoustic description, not a judgement of whether your voice is pleasant, authoritative or attractive — those are listener responses shaped by accent, content, context and familiarity, none of which any signal-processing tool can measure.

It is also not a clinical assessment. Jitter, shimmer and harmonic-to-noise ratio are genuinely used in voice clinics, but there they are measured on calibrated equipment, on a sustained vowel, under controlled conditions, and interpreted alongside a laryngeal examination. A five-second take on a laptop in your kitchen is not that. If your voice is persistently hoarse for more than two or three weeks, that is a reason to see a doctor regardless of what any website tells you.

Frequently Asked Questions

More Voice Tools