Research on Music Emotional Expression Based on Reinforcement Learning and Multimodal Information

Table 1

Different levels of audio description features.

Categories	Characteristics
LLDs	MFCC, zero crossing rate, spectral centroid, spectral bandwidth, chromaticity characteristics
HSFs	Maximum, mean, variance