Vocal range test, in this tab, in two minutes.
Sing down to your lowest note, then up to your highest. You get both notes by name, how wide your range is in octaves, and the voice type it sits closest to. No app, no account, and nothing you sing leaves your machine.
Two sweeps, about a minute each. Your browser will ask for the microphone. The audio is analysed in the page and never sent anywhere.
Step 1 of 2, your lowest note
Your result
One end of your range went past what this page can measure. The floor is a C2, and under it the detector stops giving a number it can stand behind, so your real low note is at or below the one shown. It says so rather than printing a guess.
What the microphone actually heard
Low sweep. The dashed line is the note that was kept.
High sweep. The reading is drawn raw, before any octave correction, so you can see what the detector saw.
Open the trainer, level 1 free
Knowing your range does not widen it, and it does nothing for the notes inside it. That is what the trainer is for, thirty exercises over six levels that score every attempt in cents on a named note. Level 1 is free here, all six are 9.99 once.
The first one is built and free. The other two are not, and I am working out which to do next. Clicking one of those two tells me, and does nothing else.
- The songs that actually sit inside your range, free, it exists now
- The trainer, working properly on a phone, about 5 dollars, once
- A range sheet for a whole choir, about 9 dollars, once
The monitor is the same detector, running continuously, with the cents readout and a ten second trace.
Running range checks on a whole choir, if you are measuring a room rather than yourself. Twenty singers in one rehearsal, and what to do with the twenty pairs of numbers. There is a printable record sheet to write them on, free, thirty lines.
What should I build next
Three things people keep asking me for. The first one now exists and costs nothing, 319 songs with the band their melody actually sits in. The other two are not built, and I would rather make the one you would use than guess. Clicking either of those tells me which. No payment, no email, no signup.
- The songs that actually sit inside your range, free, it exists now
- The trainer, working properly on a phone, about 5 dollars, once
- A range sheet for a whole choir, about 9 dollars, once
How to get a result you can trust
The test listens for a note you hold. A single frame is never counted, because one frame at the bottom of a sweep is almost always the detector slipping an octave rather than you singing. What it asks for is about a fifth of a second of steady pitch, so a slow slide can still leave a mark on the way past. Hold each end for a second or two and the reading settles on the note you meant.
- Sing an open ah, out loud. Humming is quieter and the low end of a hum is often too breathy to read.
- Start on a note that feels easy, then slide. Do not jump straight to the extreme, you will strain and the number will be worse.
- At the bottom, stop before it turns into a creak. Vocal fry has almost no pitch in it, and a range built on fry is a range you cannot sing in.
- Quiet room, microphone not too close. A phone held at arm's length works fine.
If you sing low, your microphone is probably lying to you. A reader tested his laptop mic in Audacity on 5 September 2026 and found the waveform simply vanishes below C2. Most laptop and headset mics are voiced for speech, which lives above roughly 150Hz, and they roll off hard underneath that. The test switches off echo cancellation, noise suppression and automatic gain, which are the three things a browser would otherwise use to flatten exactly what is being measured, but none of that puts back what the hardware never captured. So if your bottom note comes out around C2 or E2 and stops dead there, suspect the microphone before you believe the number. A phone usually does better than a laptop on the low end.
One page per voice type
If you are testing against a type you already suspect, each of the six has its own page. Same test, same detector, and the reference range for that type written out with what it does and does not settle. Bass, baritone, tenor, alto, mezzo-soprano, soprano. The vocal range chart puts all six on one axis if you want to see how far they overlap before choosing.
What the voice type actually means
The six types on the chart are conventions of repertoire, not measurements of you. They tell a choir director which part you are most likely to be comfortable singing, and they overlap heavily on purpose. If your range sits between two of them the page names both, because narrowing that down takes a teacher listening to your timbre and to where your voice changes register, and no browser tab can do it.
Range also moves. It is wider when you are warmed up, narrower first thing in the morning, and it grows with training. One measurement is a snapshot. Two measurements a month apart are far more interesting.
Where the number comes from
The pitch detector is the same one that runs the live monitor and the trainer. It is a YIN detector with a correction pass, and it used to read every note sharp, by 2.6 cents at a concert A and by nearly 30 at the bottom of a bass range. That failure and its fix are written up with the before and after measurements in my pitch detector read every note sharp.
The rule that a note has to be held is not a detail either. The first version of this page corrected octave slips by folding each reading onto the one before it, and a single stray frame then folded away two seconds of perfectly good readings. That failure, and what actually separates a detector slip from a singer changing register, is written up in an octave correction that could not tell a glitch from a soprano.
Its honest floor is 65.41 Hz, a C2. A real bass sings below that. When your sweep goes past the floor, this page tells you so instead of printing a number it cannot defend, which is the one place where it will give you a less flattering result than a test that guesses.
Nothing is uploaded
The microphone stream is analysed inside the page. There is no server, no account, no upload and no cookie for this test. Close the tab and the whole thing is gone. The only file that ever leaves your machine is the result image, and only when you click the button that makes it.