How the Vocal Range Calculator Works

The tools and what each one measures

The vocal tier test, voice Hz test, singer comparison and vocal range chart use the same building blocks. All of them are on the voice tools page.

Step 1: You allow microphone access

Nothing listens until you press a tool’s start button. The tool then calls the browser’s getUserMedia function, and the browser asks whether this page may use your microphone. If you refuse, the range tools let you pick your notes from menus instead. You can withdraw permission at any time in your browser’s site settings.

The tools ask the browser to switch off echo cancellation, noise suppression and automatic gain control. Those features are designed for calls and can bend or cut off a sung note. Some phones and browsers apply their own processing anyway, which is one reason results differ between devices. The vocal range calculator, register test and tier test release the microphone as soon as the result is ready.

Step 2: Your browser reads short slices of sound

The audio stream goes into the Web Audio API, the audio engine built into modern browsers. The tool reads the waveform many times a second in short blocks of samples. Each block is analysed and then discarded; it is not written to a file, stored or sent anywhere.

Step 3: The YIN method finds the pitch

A sung note is not a single frequency. It has a fundamental, which is the pitch you hear, plus harmonics at two, three and four times that frequency and more. A simple “loudest frequency” approach often picks a harmonic and shows a note an octave too high.

The tools use YIN, a well-established pitch detection method that works on the waveform itself. It compares each block with delayed copies of itself and looks for the smallest delay at which the wave repeats. That delay is the period of the note, and the frequency is one divided by the period. The settings used by the plugins are:

• a YIN threshold of 0.15, the level at which a repeat counts as found
• a clarity of at least 0.8, so breath, noise and rough sounds are ignored (0.75 on the voice Hz test)
• a search band of 60 to 1,400 Hz, roughly B1 to F6, which covers bass to high soprano (1,600 Hz on the register test, 1,100 Hz on the voice Hz test)

Step 4: Only steady notes count

Voices slide, wobble and crack, so a single reading is not enough. The vocal range calculator smooths the readings with a median of the last five, and a note only counts once you hold it for 0.8 seconds within one semitone. The octave range, voice type, register and tier tests and the singer comparison use the same steady-note rule. You save your lowest note first and then your highest, and you can redo either one.

Step 5: Frequency becomes a note

Once a steady frequency f is found, it is placed on the piano scale using equal temperament, with A4 = 440 Hz as the reference. The note number n (the MIDI number, where A4 is 69) is:

n = 69 + 12 × log₂(f ÷ 440)

The nearest whole number is the note; the remainder, times 100, is how many cents sharp or flat you are. Examples from the tools’ own test runs: 220 Hz is exactly A3; 100 Hz is G2, 35 cents sharp; a tone of 445.1 Hz reads as A4, 20 cents sharp. The frequency to note converter works the other way too, and lets you choose another reference such as A4 = 432, 442, 443 or 415 Hz.

Step 6: Range becomes semitones and octaves

With two notes saved, the range is simply the difference between their note numbers. Twelve semitones make one octave, so octaves = semitones ÷ 12. For example, E2 is note 40 and G4 is note 67: 27 semitones, which is 2.25 octaves or “2 octaves and 3 semitones”. The calculation is exact for the notes you saved, even though detecting those notes is an estimate.

The calculator then places your span in a band: under 12 semitones, 12 to 17, 18 to 30, 31 to 41, and 42 or more. Most untrained voices land in the 18 to 30 band, about one and a half to two and a half octaves.

Step 7: Range is compared with voice types

The site uses one set of reference ranges everywhere:

The vocal range calculator scores each type by overlap: the notes you share with that type divided by the notes covered by either range. It subtracts a small penalty for the distance between the two centres (distance in semitones ÷ 48), and if the top two scores are within 0.06 it says you sit between them. A2–A4 therefore comes out as baritone, and E2–G4 as bass. The singer comparison ranks 35 singers with the same overlap measure.

This is a match by range only. Real voice classification also listens to timbre, tessitura, weight and where the voice changes register, which no pitch tool can measure.

Registers, audio files and calculators

The vocal register test does not detect chest or head voice by itself. You choose the register and sing, and the tool measures the notes you save. It compares your highest chest note with your lowest head note: if they overlap, the shared notes are your bridge; if the head voice starts more than a semitone above the chest voice, the notes in between are a gap.

The note detector can also read an audio file. The file is decoded and analysed inside your browser (files up to 25 MB, the first 120 seconds); it is not uploaded. The frequency converter and the octave, semitone and interval calculators use no microphone at all: the same input always gives the same answer.

Where your audio goes: nowhere

The tool code contains no function that records audio or sends it over the network. Your voice is analysed in the browser tab and forgotten block by block. The tools do not save results in cookies or browser storage. A “copy link” result keeps the range after the # sign of the address, a part of the link that browsers do not send to the server. Google Analytics 4 counts page visits for the site as a whole; the tools do not send your notes or results to it.

How the site protects data in general is covered on the data security page.

How the tools are tested

Each tool version is tested before release in a WordPress test installation with Microsoft Edge. Pitch tools are fed known tones through a simulated microphone, so the right answer is known before the test starts. The checks cover calculations, the microphone path, error messages, mobile layout and structured data.

Not tested so far: Safari on Mac and iPhone, Firefox, Chrome on a physical Android phone and, for some tools, real human voices on phones. Test tones are cleaner than a real voice in a real room, so treat these results as proof that the maths and logic are right, not as a promise for every device.

Limits and how to get consistent results

• Microphone quality, distance and background noise all affect detection.
• Breathy tones, vocal fry, strong vibrato and fast slides are harder to read.
• A very strong harmonic can still cause an octave error; repeat the note to confirm it.
• Warm-up, fatigue and time of day change what you can sing.
• Use a quiet room, keep the same distance, sing one clear “ah” per note and never force a note.

Small differences between sessions are normal. For the full list of causes and fixes, read the vocal range test accuracy guide.