slaytheaccent
enter
phoneme-level coach · v0.1 iridescent build

slaytheaccent

A coach that listens at the phoneme. Record three seconds, and three acoustic models agree — or disagree — about every sound in your mouth.

phonemes44
dialects23
latency540ms
§ II · the 44 soundshover · preview
ppat
bbat
tten
dden
kkit
ggot
ffan
vvan
θthin
ðthis
ssee
zzoo
ʃshe
ʒvision
hhat
mman
nno
ŋsing
llip
ɫcool
ɹred
wwe
jyes
ɾwater
isee
ɪsit
ebed
ɛmet
æcat
ʌcup
əago
ɚbutter
utoo
ʊput
oboat
ɔthought
ɑfather
day
high
boat
now
ɔɪboy
ɝbird
ʧchip
§ III · how it works

you speak. the mouth gets measured.

phase 01capture

speak

Three seconds, the phone mic, no scripts. Silero VAD trims the edges so you don't have to.

audio → vad → window(t=3s)
phase 02score

measure

Three wav2vec2 variants emit per-frame phoneme distributions. A dynamic-programming alignment maps observed → expected.

frames → align → Δ per phoneme
phase 03drill

refine

The weakest phonemes rise to the top of the queue. You shadow the native, re-record, watch the score move.

queue.head → shadow → repeat
three weeks in and my name finally sounds like mine again.
— a. · kraków → san francisco · beta cohort 03
launch sequence

tune the instrument.

sign in with google
© 2026 · slaytheaccentiridescent build · mmxxvi