Figure 2From: Very Low Rate Scalable Speech Coding through Classified Embedded Matrix QuantizationThe graph for codewords at depth . Arc value is determined based on the encountered distortion resulting from merging th and th codewords at depth .Back to article page