Caption Shades

Mic Speech Lag — Auto · ready
Mode
Auto shows EQ after 2½ seconds of sound without words. Talk to bring captions back; EQ returns 3 seconds after the last speech result. Other room sounds can also trigger EQ.

Playing the demo. Tap Start listening, allow the microphone and speech recognition, then talk.

Your words show up here as the browser hears them.
Settings and diagnostics
16 dB How much louder than your normal voice counts as a full shout. Left = more sensitive. Right = less sensitive. This does not change EQ sensitivity.
Words
Scroll slides the line left as you talk. Pop on drops each word in place and starts a fresh line when it fills up or you pause. One word shows each word on its own.
14 bars 78 across · 1,020 LEDs · 57% see-through · 1.8 mm gaps Rows of LEDs on the grille. More bars means sharper text but less to see through.
LED size
1 mm LEDs fit 78 across for sharper text. 1.6 mm LEDs are about twice as bright but fit 66 across on wider bars. A phone screen can’t show the brightness difference.
Display
LED grille shows what the hardware could actually draw. Ideal shows the design at full resolution.
Draws words the moment the browser guesses them, slightly dimmed, then locks them in. Turn off to compare the lag.
Shows a cursor the moment you start talking, colored by how loud you are, so the glasses react before the words arrive.
Required for live EQ and loudness tracking. Turn off to test captions alone if microphone analysis and speech recognition conflict.
Calibrate your normal voice While listening, pause the music, tap and talk normally for 4 seconds. Automatic learning uses audio near recognized words; it cannot separate your voice from music playing at the same time.
Room noise—
Normal voice—
Level now—
Intensity—
Heat—
Lag (median)—
Session log Copies timings, EQ transitions and setting changes so you can share them for tuning. Auto EQ is an experimental sound fallback, not a music classifier.

If Speech shows blocked: turn on Dictation in Settings › General › Keyboard, allow the microphone for Safari, then reload. Use Safari itself, not a home-screen shortcut.