AI & ML interests

Small, offline models that find exactly when every word is spoken in a recording. CavaAlign v1 does it in 35 languages from a single 16 MB file, and powers read along in the Cava Reader app. Open under the AGPL. Interested in on-device speech, multilingual alignment, and accessibility.

Recent Activity

micr0-dev  published a model 4 days ago
cava-reader/cavaalign-v1
micr0-dev  updated a model 4 days ago
cava-reader/cavaalign-v1
micr0-dev  updated a model 7 days ago
cava-reader/cavaalign-v1
View all activity