Why this tool exists
Recorded voices often sound different from the voice we hear while speaking. This project turns that curiosity into a quick, approachable test for speakers, performers, streamers, podcasters, and anyone comparing how a voice or microphone sounds.
How it is built
The test runs in the browser using MediaRecorder and the Web Audio API. It measures fundamental frequency, pitch variation, harmonicity, and spectral centroid. The current version does not analyze formants. Audio stays on the device during analysis and is not uploaded to our server.
What we believe
A voice cannot prove gender identity, biological sex, actual age, health, or authenticity. Our labels describe a sample’s acoustic impression, not the person speaking. We publish limitations prominently and do not position the result as professional guidance.
Editorial standard
Explanations are written to match the behavior of the current tool. We distinguish measured features from broad tendencies, avoid claiming certainty, and update documentation when the implementation materially changes. Read the detailed guide to how voice gender detection works.
Try it for yourself
Record a natural three-to-five-second sample. No registration is required.
Open Voice Gender Detector