Rate My Voice

Rate My Voice in Three Steps

The whole point of a rate-my-voice test is to get an answer before you lose interest. There is no account, no email field and no upload step, so the slowest part is finding a quiet corner.

1. Press record. Hold your phone or sit about a hand-span from your laptop microphone, at the distance you would normally use on a call.
2. Read a full sentence out loud for five seconds, at your everyday speaking volume. Reading works better than humming or counting, because the test needs real speech to measure clarity and delivery.
3. Read the breakdown. You get one number from 1 to 10, plus the seven sub-scores that produced it, so you can see which part of your voice pulled the score up or down.

Retakes are unlimited. Most people record two or three times before the number settles, which is normal: the first take is usually the quietest.

Rate My Voice From 1 to 10: Where Your Score Lands

These bands come from calibrating the engine against reference values for seven voice types, from trained speakers through to hoarse and strained voices. They tell you what a given number usually means.

• 9.0 and above: voice-over territory. Clear, steady and well projected, almost always recorded in a quiet room.
• 8.0 to 8.9: a healthy, clear voice with good control. This is where a confident presenter usually lands.
• 7.0 to 7.9: the ordinary adult range on a laptop or phone microphone. Nothing wrong; nothing polished either.
• 6.0 to 6.9: a tired or slightly strained voice, or a decent voice recorded too far from the microphone.
• 5.0 to 5.9: hoarseness, a cold, or noticeable background noise.
• Below 5.0: usually a recording problem rather than a voice problem. Move somewhere quieter and record again before drawing conclusions.

The word under your number

The result card prints a one-word grade under the score out of 10. These are the exact cut-offs the page uses:

• 9 out of 10 or higher: Outstanding
• 8 to 8.9: Very strong
• 7 to 7.9: Solid
• 6 to 6.9: Decent
• 4.5 to 5.9: Rough around the edges
• Below 4.5: Needs work

So a 7 out of 10 is not a weak result. It is the middle of the normal range for an adult speaking into a laptop or phone, and it earns the Solid label. Most people who want a higher number get there faster by fixing the room and the microphone distance than by changing how they speak.

If you want to rate your singing voice instead, the same engine with different weights is on the rate my singing page.

What the Rate My Voice Test Measures

The single number at the top is a weighted combination of seven separate measurements. Each one deserves a look 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.

Example result · sample recording, not your voice

Clarity: 32% of the score

This is smoothed cepstral peak prominence, or CPPS: how closely the spectrum of your voice resembles a clean harmonic comb instead of a smear of breath, turbulence and room noise. It carries the heaviest weight because it is what listeners consciously notice first. Here the sub-score is zero at 3.2 dB and full at 9.5 dB. In our own tests, healthy adults reading a passage aloud measure about 6.7 to 7.0 dB, and a poor recording of the same voice can drop to 3.1 dB. Absolute CPPS values differ between implementations, so these figures are calibrated on this tool's own measurements and do not transfer to Praat or clinical systems. Nine times out of ten a low reading is the room, and the larynx is fine.

Pitch stability: 16%

Measured as jitter: the percentage by which each vocal fold vibration cycle differs in length from the one before it. The measurement is taken cycle by cycle in the time domain. Averaging frame by frame across 40 milliseconds would smooth out the very variation the test is looking for. The familiar 1% threshold belongs to a sustained vowel and does not survive the move to connected speech, where ordinary intonation and word transitions are counted as perturbation: real speakers reading aloud measure between 2.5% and 8.2%. The scale is set for that, with full marks at 2% and zero at 7%, and the weight is modest because on running speech jitter says more about your intonation than about your vocal folds.

Volume control: 12%

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. Connected speech typically measures 1.0 to 2.4 dB, so full marks sit at 0.8 dB and the sub-score reaches zero at 2.8 dB.

Expressiveness: 14%

How far your pitch moves around its own median, measured in semitones. 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 signal instability, so this sub-score peaks in the middle of the range and falls off above it.

Tone brightness: 12%

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 often comes from microphone position. Speaking across a laptop lid, or into a phone lying face down on a desk, drags this figure down sharply.

Delivery flow: 8%

The proportion of your take that carried full voice. Breathy, whispered or silent stretches do not count. 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 does not need a wide range. It is weighted far more heavily on the singing test.

How to Improve Each Sub-Score

Two of these sub-scores respond to your recording setup within seconds, and the rest respond to practice over weeks. Know 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 instead of straight 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. A low first result usually says more about your setup than about 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, so practice changes them. Ten minutes a day of reading aloud moves both over a month. Vary pace and pitch on purpose, and finish each sentence with support instead of letting it trail into breath. This is the area where recording yourself weekly and keeping the numbers pays off.

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 Change Your Score

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 does account for this. It measures the noise floor in the silent frames between your words and compares it with your speaking level, treating 12 dB of signal-to-noise as a dirty recording and 34 dB as a clean one. A noisy take then has up to 3.2 dB added back to its clarity figure, so a bad room is not charged twice against the voice. The compensation is partial by design, and a quiet room still beats it.

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. 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.

How much should a score move between takes? We split one 20-second reading into seven 5-second takes and scored each one: the same speaker landed anywhere from 7.0 to 8.3. The full test and a simple three-take method are in why your voice score changes every time you test.

How to Read the Score Fairly

The score is an acoustic description. It does not judge whether your voice is pleasant, authoritative or attractive. Those are listener responses shaped by accent, content, context and familiarity, and no signal-processing tool can measure them.

Put plainly: when people ask a tool to rate my voice, they usually want to know whether they sound good to other people. Acoustics can only answer half of that question. The score tells you how clear and steady the signal is, not whether your delivery is persuasive or your accent suits a given audience.

It is also not a clinical assessment. Voice clinics do use cepstral peak prominence, jitter and shimmer, but they measure them on calibrated equipment, on a sustained vowel, under controlled conditions, and interpret them 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.

Measured on Real Recordings: How Much a Score Moves

We scored three public-domain LibriVox readings with the same code this page runs, first as one 20-second take and then as seven overlapping five-second takes.

The same voice moved by up to 1.3 points between takes, while the rough recording stayed below the two healthy voices every time. Compare takes recorded the same way, not single numbers. The full test is in why your voice score changes.

Frequently Asked Questions

More Voice Tools

Same engine, different questions: rate my voice covers everyday speech, while the tools below look at singing, perceived age and perceived gender.