SingingRangeTest.com offers free browser tools that measure pitch and estimate things such as voice type. They are useful for exploring your voice, but they are not laboratory instruments. This page says how accurate they can be expected to be, what has and has not been tested, and what can go wrong.
The short answer: we do not publish an accuracy percentage, because none of the tools has been tested on real microphones or real voices. What we can share is what was tested with generated signals, the errors that showed up, and the conditions that change a result. The rules behind each tool are in the Testing Methodology.
What has been tested, and what has not
The ten rebuilt tools were tested with computer-generated tones, noise and speech-like signals, and with simulated browsers and a fake microphone, on one Windows computer on 2026-09-29 and 2026-09-30. Nothing has been tested with:
- real people singing or speaking into real microphones;
- Safari (Mac or iPhone), Firefox, Android, or any real phone or tablet;
- Bluetooth or USB microphones and headsets;
- a real WordPress installation, or a screen reader.
The homepage range test has not been through this kind of review. The Pitch Accuracy Test has no test record on file, only a code reading. A tool that works in a test browser can still behave differently on your phone, particularly when it first asks to use the microphone.
What the generated-signal tests showed
These results describe generated signals, not human voices. Real voices are less steady and noisier.
| Situation | What the tests found |
|---|---|
| Clean generated tones | The pitch detectors were within about 2 cents or better in every tool’s own tests (worst case 2.1 cents). |
| Tone plus noise | In noise about 10 dB below the tone, readings stayed within about 10 cents with no octave errors. In very heavy noise (6–12 dB), the High Note Test read about 2% of frames wrong and about 23% not at all. Noise alone was never read as a pitch in the tests. |
| Octave slips | Sets of deliberate octave traps (weak or missing fundamental, strong harmonics) produced no octave errors in the Singing Note Detector or Note Finder tests. The High Note Test still misread very weak low fundamentals an octave up in 0.2–0.6% of frames, and Note Finder about 1 frame in 3,000. |
| More than one note | All tools assume one note at a time. On generated chords, C4 + E4 was read as C2, and C4 + G4 as C3. |
| Voice Quality Test | On generated voices, the jitter-like readout was 0.53–1.16 times the true value and the shimmer-like readout 0.71–0.79 times. Its numbers are not interchangeable with Praat or clinical reports. |
| Singing Grader | Steady random pitches (not a song) still graded about 7 out of 10, because the grid fit makes centring less discriminating. Very wide vibrato (about ±100 cents) can split one note into several. |
| Vocal Scale Finder | The right scale ranked first in 72 of 72 generated scale-shaped phrases, but real melodies use fewer notes: four notes can fit nine scales, and a relative major and minor share every note. |
Why real-world results differ
Microphone and device processing
Phone, laptop, headset and USB microphones differ in sensitivity, frequency response and built-in processing. Our tools ask the browser to switch off echo cancellation, noise suppression and automatic gain, but a browser or device may ignore the request, and this has not been tested on real devices. The Voice Quality Test shows what your browser reports. Low-rate Bluetooth headsets can smooth the signal. An external microphone is not automatically more accurate than a built-in one.
Background sound and room
Pitch detection works best when your voice is the clearest pitched sound. Music, television, another voice, fans and echo can make readings unstable or wrong. The Voice Quality Test flags when room noise limits its clarity reading.
Octave errors and unstable notes
A detector can occasionally take a harmonic for the fundamental and show a note an octave away. The tools use octave guards, median smoothing and hold times to limit this, but they cannot remove it. Slides, scoops, strong vibrato, breathy onsets, whisper and vocal fry can make a note hard to pin down; several tools show “no clear note” or similar instead of guessing. If a note looks obviously wrong, repeat it.
Your voice on the day
Warm-up, tiredness, illness and how you sing (softly, pushed, breathy) change what you produce. There is no single warm-up length or time of day that guarantees the best result; consistency matters more. A narrower result on a tired day does not mean lost range, and a wider one is not proof of progress.
Tools cannot feel strain
A tool can tell that a pitch was stable, not whether it was comfortable, strained or healthy. Do not force a note to widen your result, and stop if it hurts. The U.S. National Institute on Deafness and Other Communication Disorders advises avoiding speaking or singing when your voice is hoarse or tired: Taking Care of Your Voice. Persistent pain, recurring voice loss or unexplained hoarseness call for a qualified healthcare professional; see also our vocal health tips.
How to get more consistent results
- Use a quiet room and switch off music and television.
- Use the same device, microphone position and test mode when comparing sessions.
- Sing clear, steady notes rather than slides; avoid moving the microphone.
- Repeat any surprising reading before believing it.
- Open the site in your normal browser if an in-app browser (from social media, for example) will not start the microphone.
- Test when your voice feels normal for you, and compare several sessions rather than one.
Reading an unexpected result
- Range looks too wide: suspect an octave slip, background sound or one isolated extreme note. Repeat it in a quiet room.
- Range looks too narrow: the input may have been weak, some notes may not have been held long enough for the tool’s hold rule, or your voice felt different that day.
- One note looks about an octave off: repeat it; an octave error is the most likely cause.
- Voice type looks wrong: the three-note voice type estimate compares your notes with fixed textbook ranges and cannot hear timbre, weight or register changes. Treat it as a starting point.
What our tools do not measure
- Singing ability, rhythm or tone. The Singing Grader scores pitch centring, steadiness, clear-pitch time and range only. It does not score rhythm, timing, tone or whether you sang the right notes of a song.
- Vocal health. None of the tools inspects the vocal folds or diagnoses anything, and the Voice Quality Test gives no health, age, gender or identity reading.
- True comfort or tessitura. The comfortable zone in the voice type estimate follows a fixed formula from your lowest and highest notes.
- Your register or passaggio. Pitch alone cannot show which register produced a note; register-change notes are reference text.
- Long-term potential. A test shows one session, not your maximum future range.
When not to use a result
Do not use a result as a medical diagnosis, proof that your voice is healthy, a professional certification, a definitive classical Fach, a reason to force repertoire, or proof of permanent range gain or loss. The tools are for education and exploration.
Try the tools and read more
- Pitch Accuracy Test
Sing target notes and see how many cents off you are.
- Singing Grader
A 1 to 10 grade for a short take, with the formula shown.
- Climb to your top note
Hold each rising target for one second and see the highest you clear.
- Voice Frequency Test
Your speaking or singing pitch in Hz.
- Live note readout
The note you are singing right now, with cents and a log of held notes.
- Voice type from three notes
An estimate from your lowest, highest and easiest note.
- Note Finder
Notes in order on a staff, from a phrase or an audio file.
- Voice Quality Test
Separate readouts for one held vowel, with no single score.
- Which scale am I singing?
Ranks 72 scales against the notes you hold.
- Vocal Pitch Changer
Shift your own recording up or down and save it.
Continue with Testing Methodology for the exact rules, How the Singing Range Test Works for a walk-through of the main test, the FAQ for quick answers, and the Privacy Policy for data handling. Sourcing and corrections are covered in the Editorial Guidelines.
SingingRangeTest.com is created and maintained by Sam Cooke; see the Author page. To report an error on this page, use Contact Us.
Page last reviewed: .