Weeks 9-10 · transfer · spoken-to-sung contour
Glide from spoken to sung without your voice breaking.
The rule
pass: ≤ 35¢ median contour deviation in 4 of 5
Two things have to be true at once. Top panel: the pitch contour you spoke must come back when you sing — same up-down shape, just lifted onto a steady pitch. Bottom panel: voicing stays unbroken across the transition. Grey shows the failure modes — a pitch jump (wrong note on the sung side) or a glottal break (gap in voicing). Either one fails.
Hear it first
Three references. The clean transfer keeps the same melodic shape on both sides. The pitch jump arrives on the song side at a totally different note. The voice break re-onsets with a noticeable click — same contour, but the listener heard you restart.
What’s happening. Spoken English already has pitch — most people inflect a melody of 100-300 cents across a sentence without thinking. Singing the same words requires lifting the larynx into steady-tone mode without losing that contour. The audio-motor loop trained in single-note-match has to extend to a moving target. Untrained singers typically deviate 80-150 cents when they try; trained singers fall under 35.
Why 35¢ tolerance on contour, not absolute pitch.We don’t care which key you sing in — we compare each sample to your own spoken median, so the score measures shape match, not pitch memory. Thirty-five cents is roughly a quarter-tone, the line where listeners stop hearing “same melody” and start hearing “different melody.” The voicing-continuity check is binary: any unintended silence in the bridge fails the trial.
Why this is the bridge.Every transfer-family lesson after this one — phrase-from-memory, song-from-memory, a-cappella-resolve-tonic — assumes you can move audio that’s in your head into a sung phrase without inventing a new pitch on the way out. Get this connection settled here so the harder lessons aren’t doing two jobs at once.
Session ahead
5 short motifs · speak 2.2s, sing 2.8s · 4/5 inside the 35¢ band promotes.