Analyze
Watch the shape of your voice as you sing it — the harmonics, the ring, and how much work your folds have actually done today.
Nothing is uploaded. The microphone is read in your browser, the analysis runs on your device, and no audio leaves it.
Time runs left to right, pitch bottom to top. Stacked lines are the harmonics of one note; the stack jumping is a register change, and a steady ripple is vibrato.
Note —
The gold column is the singer’s formant, the cluster around 3 kHz that lets a trained voice sit on top of an orchestra playing louder than it. The percentage is that band’s share of the plotted energy — compare it against your own takes rather than against a target.
- 2026-08-17: 0s voiced, 0 cycles
- 2026-08-18: 0s voiced, 0 cycles
- 2026-08-19: 0s voiced, 0 cycles
- 2026-08-20: 0s voiced, 0 cycles
- 2026-08-21: 0s voiced, 0 cycles
- 2026-08-22: 0s voiced, 0 cycles
- 2026-08-23: 0s voiced, 0 cycles
Cycle dose counts how many times your vocal folds opened and closed — pitch multiplied by the time you were actually sounding. An hour spent high in your range is far more work than an hour spent low, which is why a stopwatch alone tends to understate a soprano’s day.
Time spent listening counts toward your practice XP — sessions log automatically every 5 minutes and when you leave the page.
What a spectrogram shows a singer
A spectrogram plots every frequency in your voice against time, so one sung note appears as a stack of horizontal lines: the fundamental at the bottom and its harmonics above. It shows the parts of singing you cannot hear from the inside — where a register changes, how fast a vibrato moves, and how much energy sits in the band that makes a voice carry.
Pitch tells you which note. Everything else about how a voice sounds — bright or dark, thin or full, ringing or swallowed — comes from the relative strength of the harmonics above that note, and that is what a spectrogram makes visible. Two singers on the same pitch produce completely different pictures.
The band around 3 kHz is worth knowing about. Trained classical voices tend to show a cluster of energy there, usually called the singer's formant, and it is the accepted explanation for how one voice is heard over an orchestra that is measurably louder. It is not a target to chase directly; it tends to appear as a by-product of an efficient, well-resonated tone.
Vocal load is a different question from how long you practiced. The vocal-dosimetry literature counts vibration cycles rather than minutes, because pitch decides how much work the folds do in a given second: a soprano rehearsing high for an hour has put her folds through several times the cycles of an hour spent low in the range. Counting cycles is what makes two practice days comparable.
How to read your voice on a spectrogram
The analysis runs in your browser on your own device. Nothing is uploaded, and nothing is recorded unless you use the take recorder separately.
- 01
Allow the microphone and sing a steady note
Pick a comfortable note in the middle of your range and hold it on an open vowel. You should see a stack of horizontal lines: the lowest is the note you are singing, and the ones above it are its harmonics.
- 02
Count how far the stack goes
A bright, well-supported tone shows many harmonics reaching high up the display. A breathy or pressed tone shows fewer, and more of the fuzzy grey between them, which is air rather than tone.
- 03
Slide slowly from low to high
Watch the whole stack rise together. Where the picture suddenly reorganises — harmonics dropping out or the texture changing — is a register transition, and it usually happens in the same place every time.
- 04
Hold a note with vibrato
Vibrato appears as a regular ripple in every line at once. Because all the harmonics move together, a wide ripple low in the stack means a wide one high up too.
- 05
Watch the ring band on the tone view
The gold column marks 2.8 to 3.2 kHz. Sing the same phrase with different amounts of space and brightness and watch the percentage move. Compare it against your own takes rather than against any fixed number, because it depends on your microphone and how close you are to it.
- 06
Let vocal load run through a normal session
Leave the page listening while you practice. It counts only the time your voice is actually sounding, so silences and talking between exercises do not inflate it.
Look at the shape, not the numbers
Early on the useful thing is noticing that the picture changes when you change something. You do not need to interpret a spectrogram to benefit from seeing that a supported note and a breathy one look nothing alike.
- Sing one note two ways — breathy, then firm — and watch how much of the stack fills in
- Find where your register changes by sliding slowly, and remember roughly which note it happens on
- Ignore the ring percentage at first; the spectrogram picture is more useful to a new singer
- Use vocal load to see whether you actually practiced twenty minutes or four
Use it to compare, not to score
Every reading here is relative to your microphone, your distance from it, and your room. That makes it excellent for comparing two of your own takes recorded the same way, and close to meaningless as an absolute measurement against another singer.
- A/B two vowel shapes on the same pitch and watch the ring band rather than trusting the sensation
- Track cycle dose across a heavy rehearsal week — the number climbs faster than practice minutes do
- Watch the harmonic stack through a passaggio to see whether the transition is smoothing out over months
- Keep microphone position consistent between sessions or the comparison is not one
What is a spectrogram in singing?
It is a display of every frequency present in your voice, plotted against time. One sung note shows as a stack of lines — the fundamental plus its harmonics — and the pattern of those lines is what makes a voice sound bright, dark, full or thin.
What is the singer's formant?
A concentration of energy around 3 kHz found in many trained classical voices. It is the standard explanation for how a single unamplified singer is heard over a full orchestra, since the ear is very sensitive in that region and orchestral sound has comparatively little there.
What is vocal dose and why count cycles instead of minutes?
Vocal dose measures how much work the vocal folds have done. Cycle dose counts vibration cycles — pitch multiplied by the time you were actually voicing — because a minute sung high puts the folds through far more cycles than a minute sung low. Minutes alone treat those as equal.
Does this measure how loud I am in decibels?
No. A browser cannot know the sensitivity of your microphone or how far your mouth is from it, so the levels here are relative to your own setup. They are useful for comparing your takes with each other, not for measuring sound pressure.
Can it tell me which vowel I am singing?
Not currently. Vowel identity comes from the first two resonances of the vocal tract, and at high pitch the harmonics are spaced too far apart to locate them reliably. The spectrogram shows the resonance structure; it does not label it.
Is my audio uploaded anywhere?
No. The microphone is read in the browser and the analysis runs on your device. No audio is sent to a server.