Speech emotion recognition (SER) is a technique for accurately determining a person’s emotion from their speech. Enhancing communication between people and machines. Even though emotions are arbitrary, it can be hard to annotate audio and it’s still hard to gauge someone’s emotional state. It now plays a crucial part in technological research as new technologies for human machine or medical applications are developed. It is obvious that surprise and rage feelings are more accurately predicted by our model than other emotions, It is understandable given how significantly the pitch, speed, and other qualities of the audio files for these emotions differ from those of other audio files. The system can capture and analyze various acoustic elements contained in voice signals using the SVM-based architecture and enabling the identification and categorization of various moods. Mel frequency cepstral coefficients (MFCCs), pitch, energy, and spectral features are some of these features, and they all work together to reveal important details about the emotional content that is present in speech. The deployed models attain a remarkable accuracy of 80% in identifying various emotional states from speech data after extensive training and optimization. This level of accuracy shows how well the suggested method works in properly classifying emotions based on auditory characteristics.