In applications involving human-machine interaction, emotion recognition from speech signals and deep fake detection are broad and significant research areas. This paper focuses on the exploration of various technologies, particularly classifiers, for recognizing emotions embedded in speech signals and detecting deep fakes. The study involves classifying speech signals into seven distinct emotional categories: happiness, fear, pleasant surprise, sadness, anger, neutral, and disgust. For deep fake detection, Mel-Frequency Cepstral Coefficients or MFCC is being used for extraction of features. The classification performance is evaluated using a Long Short-Term Memory or LSTM network for the speech emotion recognition and an ensemble of speech emotion recognition (ESER) outputs for deep fake detection.