how to reduce overfitting and improve confusion matrix

I am trying to apply the following model on my data which is consists of (4030 samples as 5 classes) each sample is MFCC features which is extracted from an audio clip consisting of (20 second) and I am trying to apply classification, but I got very poor accuracy and I also have overfitting, , Although I am using data augmentation and I also try to apply Batch Normalization to improve overfitting but the result is very bad.

the Model:

Effnet=tensorflow.keras.applications.EfficientNetB7( input_shape=(IMG_SIZE,IMG_SIZE,3), 
include_top=False,weights=imagenet,pooling=avg)
Effnet.trainable = False
x = Flatten()(Effnet.output)
x=(BatchNormalization())(x)
#add two fully connected dense layers 1024 as my model 
x=Dense(1024)(x)
x=(BatchNormalization())(x)
x=Activation('relu')(x)
x=Dense(1024)(x)
x=(BatchNormalization())(x)
x=Activation('relu')(x)
x = Dense(NUM_CLASSE)(x)
x=(BatchNormalization())(x)
prediction =Activation('softmax')(x)
model = Model(inputs=Effnet.input, outputs=prediction)
model.summary()

the learning curve: the confuusion matrix:

Any help, Regards in advance!