Taha ChellyQuranic ShazamWhen an audio match needs more context
Separating speaker recognition from verse discovery, and making uncertainty part of the listening flow.
Taha Chelly
Engineering notebook
Technical decisions, explained.
Taha ChellyQuranic ShazamTaha ChellyQafiyaArabic prosody on a small CPU server
A 4.8 MB vocalization model, preserved vowel marks, and explicit limits on what automated analysis can claim.
