Add the 8-bit encoder as an unzipped folder (per-file downloads)

Add 8-bit palettized Core ML encoder for whisper large-v3-turbo

initial commit