Weeks 5-6 · voice · vowel transition stability
Change the mouth shape. Keep the pitch dead flat.
The rule
pass: pitch SD ≤ 30¢ across 3 s · both vowels classified
Three seconds, one breath. The emerald trace stays inside the ±30¢ band across the vowel switch at 1.5 s — that’s pass. The grey trace dips ~50¢ right at the switch — the failure mode. It happens when the jaw drops with the open vowel and drags the larynx down with it. The colored row below the trace shows the classifier committing to vowel A then vowel B at the right moments.
Hear it first
First clip: /i/ → /a/ with rock-steady pitch — what you’re aiming at. Second clip: the same switch but pitch sags a quarter-tone on the /a/. Third: the opposite — pitch sharpens on the close vowel. Both are the same disease.
What’s happening. Vowel changes are jaw + tongue events; pitch changes are larynx events. Beginners involuntarily couple them — when the jaw drops to open the vowel, the strap muscles pull the larynx down with it and pitch sags. The drill is to decouple the two systems. Trained singers keep pitch SD inside 20¢ across a vowel switch; untrained singers commonly see 60–80¢ swings.
Why 1.5 + 1.5 s. Each phase needs about a second of settled steady-state for the spectrum classifier to commit (after ~300 ms of onset noise). 1.5 seconds is the shortest phase length where we can measure both vowels cleanly and still keep the whole drill in a single comfortable breath. The ±30¢ tolerance is the same one we use on the /u/ sustain — same line of acceptable pitch wobble, but stress-tested with a mouth-shape transition.
Why this is the bridge.Every legato lesson downstream — two-syllable-legato, consonant-vowel-clarity, lyric singing — assumes you can change the vowel without disturbing the pitch. If the jaw and larynx aren’t decoupled here, every text-bearing phrase later will wobble at every vowel boundary.
Session ahead
First-time setup: calibrate your five vowels before the session starts.