Emotion Detection Speech using LSTM neural network
feature extraction from voices using liberosa library
and Keras LSTM model for Classification
reference: A Large-Scale Validated Database for Persian Speech Emotion Detection https://arxiv.org/abs/1906.01155 dataset: Sharif Emotional Speech Database https://github.com/mansourehk/ShEMO
classes are:
N -> Natural
H -> Happy
S -> Sad
W -> Worry
A -> Angry
final accuracy: 67%