Last updated: 22 November 2026.
SingingRangeTest.com separates measured input, calculations and interpretation. This page explains the current range workflow and selected related tools. A result should be read with the task, input mode, reference tuning and recording conditions that produced it.
Capturing or entering singing endpoints
The main range tool can capture stable low and high microphone notes or accept notes you select manually. Microphone analysis uses the McLeod pitch method with clarity and stable-note filtering; a capture requires a note held for approximately 0.7 seconds. It works on one clear pitch at a time, not a polyphonic performance.
Harmonics, noise, breathy or irregular sounds and microphone processing can produce missed detections or octave errors. Inspect the note and octave before saving it. A stable detection does not establish that the note is physically comfortable. Manual mode calculates from your choices without verifying a sung performance.
Semitones, octaves and tuning
Range span is the high note’s MIDI number minus the low note’s MIDI number. Divide that distance by 12 to obtain octaves. For example, A2 to F4 is MIDI 45 to 65: a 20-semitone interval, or approximately 1.67 octaves. Counting both endpoints gives 21 chromatic note positions, which is different from the distance.
Equal-tempered frequency follows f = A4 × 2^((m − 69)/12), where m is the MIDI note number. A4 defaults to 440 Hz. Raising a note by one octave doubles its frequency; matching notes between sessions requires a consistent tuning reference.
Illustrative voice-range comparisons
The current range references are Bass E2–E4, Baritone A2–A4, Tenor C3–C5, Countertenor E3–E5, Contralto F3–F5, Mezzo-soprano A3–A5 and Soprano C4–C6. These are overlapping reference intervals, not universally definitive boundaries.
Span match compares the shared semitone interval with the union of the intervals. The percentage describes that arithmetic, not a classification probability. Endpoints alone do not assess tessitura, tone, register transitions or technique.
Singing Grader: three separate components
The Singing Grader uses a supplied comfortable range and five target-note attempts. It is not a range-only score. The target positions are approximately 20%, 35%, 50%, 65% and 80% through the supplied span, rounded to notes.
- Range: up to 40 points. Points follow the app’s defined interval bands.
- Target-pitch accuracy: up to 40 points. This uses mean absolute cents error from the target notes.
- Stability: up to 20 points. This uses the measured pitch spread during attempts.
Each attempt can run for up to eight seconds and needs at least one voiced second to be valid. A total requires all five targets to be completed or skipped and at least one valid attempt. Skipped or undetected attempts receive formula penalties; those penalties are not measurements of the person’s actual pitch.
The /100 total and tiers 1–5 are app-defined practice feedback, not population percentiles, a professional examination or a validated talent rating. Range width alone does not establish better singing. Read the grader page for the detailed thresholds and score rules.
Voice Quality Test uses a different formula
The Voice Quality Test measures fundamental pitch, spectral balance, periodicity and digital level locally. Its optional held-note consistency score is not Singing Grader divided by ten.
Speech mode has no quality score. After at least three detected voiced seconds in held-note mode, the custom score is:
10 × [0.60 × max(0, 1 − D/100) + 0.30 × V + 0.10 × (1 − C)]
D is median absolute pitch deviation around the sample’s median in cents, V is voiced-time fraction and C is clipping-frame fraction. The result is rounded to one decimal place. Natural vibrato can lower it, and a steady wrong melody note can score highly because the formula has no melody target.
These are local signal measurements, not an AI judge or clinical voice examination. Spectral centroid does not measure attractiveness; detector periodicity is not clinical HNR; digital dBFS is not calibrated acoustic SPL.
High notes, target notes and transformed recordings
The High Note Test uses the two-endpoint range workflow. Save comfortable repeatable notes rather than forcing a maximum. Live note detectors estimate one pitch at a time; target exercises compare it with the selected note, not an entire performance’s expression or rhythm.
The Vocal Pitch Changer instead transforms a selected local recording and offers WAV export. Processing can introduce audible artifacts. Transposing a recording does not establish that a singer can comfortably perform it.
Meaningful comparisons and validation limits
Repeat a comparable task with the same microphone, distance, gain and reference tuning. Inspect outliers rather than treating every detected extreme as progress. An error or unavailable result is not a measured zero.
Known numerical inputs can check interval calculations. Generated audio can check selected signal-processing cases, and browser tests can check controls in a stated environment. Those checks are different from validation with real singers, physical devices or clinical populations. We do not claim universal accuracy or independent clinical certification.
Processing, privacy and reporting
Local audio processing is separate from website delivery and any analytics services. Ezoic advertising is currently inactive on this website. The Privacy Policy explains the current services, browser storage and available privacy choices.
Send a reproducible issue to support@singingrangetest.com, including the page, input mode, notes or settings, browser and unexpected result. Prefer a description or non-identifying test input to a private recording. See the Contact page for details.
