Audio stays on your device
Microphone samples are processed in this browser tab. Saving keeps a numeric summary locally; sharing is optional. See the privacy policy for other website services.
The free online voicecel test
Measure your speaking pitch with a ten-second sample. See the median frequency and variation, then learn what those numbers mean. No account or audio upload is needed.
Your voice frequency will appear here.
100% client-side. Audio is never uploaded. Results are stored only when you choose Save.
An estimate, with limits
Background noise, vocal fry and microphone processing can affect pitch detection. We have not established an accuracy tolerance against laboratory equipment.
Statistics you can interpret
Compare median, mean, detected range and pitch variation. They describe the captured sample; they do not rate your voice or identify you.
What does a voice frequency test measure?
Fundamental frequency describes how often a voiced sound repeats each second, in hertz (Hz). This tool estimates that pattern during speech. Pauses and unvoiced sounds do not have a stable pitch, so the result summarizes detected voiced frames rather than every sound you make.
Make comparisons repeatable
Use a quiet room and keep the microphone in the same position. Read the same passage naturally for each ten-second test. Repeat three times, comparing the medians and detected sample counts. Keep your volume comfortable; forcing your pitch down makes the comparison less useful.
Example: a change is not automatically progress
Suppose three takes show medians of 120, 124 and 118 Hz. These illustrative numbers span 6 Hz; they are not study data. A later 122 Hz reading fits that spread. Compare several recordings under similar conditions before treating a difference as a lasting change.
What the result cannot tell you
A frequency reading cannot determine attractiveness, gender identity, personality or vocal health. Two voices with the same pitch can sound different because of resonance and articulation. The interface bands are descriptive display choices, not clinical limits or population percentiles.
When to repeat a recording
Repeat if few pitch samples were detected, another speaker was nearby, or the trace jumps to roughly double or half its previous value. Those jumps can be tracking errors. A narrow spread alone does not prove accuracy: natural speech includes intonation. Do not strain to obtain a preferred number.