Figure 2From: Compact Acoustic Models for Embedded Speech Recognition Process to obtain the state-independent GMM. Back to article page