Sound Classification Based on Multihead Attention and Support Vector Machine

Table 10

Comparison of the classification accuracy with six methods on IEMOCAP.

Methods	Preprocessing	Accuracy (%)
LSTM-RNN [36]	Mel-spectrogram	64.8
FAF [37]	Mel-spectrogram	61.4
HSF-CRNN [38]	Mel-spectrogram, LLDs, etc.	60.4
CNN-LSTM-DNN [39]	Raw speech	60.2
Progressive net [40]	Mel-spectrogram, MFCC, etc.	58.1
MhaNN-SVM	Mel-spectrogram, MFCC, etc.	62.8