1 paper
Soohyun Kim, Jiwon Jeong, Marina Bosi
We present a perceptual audio coder in which the entropy-coding model and the bit allocation co-adapt across blocks without model side information. Mantissas are coded by a hybrid…