Speech Clarity Tester
Read a few sentences into your mic. The tool builds your long-term spectral balance across speech bands and a consonant-presence score — a proxy for how crisp your consonants are — with practical tips.
ℹ This measures spectral balance, not true intelligibility. A real intelligibility metric (like STI) needs reference signals and room measurement; the “clarity” here is the energy in the 2.5–6 kHz consonant band relative to the rest of your speech — useful for spotting a muffled or harsh balance, but affected by your mic and room and not a clinical or pass/fail test. Your mic is analyzed live and never recorded or uploaded.
How It Works
While you speak, the tool builds a long-term average spectrum — the average energy at each frequency over the whole sample (only counting moments when you’re actually talking). The chart shows the speech bands: low-mid (vowel warmth), mid (vowel core), presence (2.5–6 kHz — where consonant detail lives), and air (above 6 kHz), setting aside the sub-300 Hz rumble/body region. Crisp, clear speech carries a healthy share of energy in the presence band, because that’s where fricatives (s, f, sh), plosive bursts (t, k, p) and the edges of words sit. The score is the presence band’s share of that speech energy, scaled to 0–100 — so it matches the cyan Presence bar. Everything runs live on your device; nothing is recorded.
It’s a useful diagnostic for a muffled balance (too little presence — soft consonants, hard to follow on calls) or a harsh/sibilant one (too much). But it is not a true intelligibility score: real speech clarity depends on articulation, pacing, background noise, and the listener, and proper measurement (STI/STIPA) uses calibrated reference signals. Your mic’s own frequency response also colours the result — compare changes on the same mic rather than chasing an absolute number.