Published Paper Details:

Emotion recognition from speech is an essential task in humancomputer interaction, with applications in mental health monitoring, virtual assistants, and customer service automation. This project, "Emotion Recognition Using Speech Processing," aims to enhance the accuracy of emotion classification by leveraging advanced speech signal processing techniques and deep learning models. The system utilizes Mel-Frequency Cepstral Coefficients (MFCCs) to extract meaningful features from speech signals, as MFCCs effectively capture the timbral and phonetic characteristics of human voice. To improve the robustness of the feature set, Principal Component Analysis (PCA) is applied to reduce dimensionality and remove redundant information, ensuring computational efficiency while retaining critical data. Additionally, an Isolation Forest algorithm is employed for anomaly detection and noise reduction, enhancing the quality of input features. For classification, a Convolutional Neural Network (CNN) is designed to learn spatial hierarchies of features, capturing intricate patterns in speech signals that are indicative of different emotional states. The CNN model is trained and evaluated using standard emotional speech datasets, with performance metrics such as accuracy, precision, recall, and F1-score used to assess effectiveness. The proposed approach aims to outperform traditional machine learning models by improving generalization and adaptability to diverse speech variations. Experimental results demonstrate that the integration of feature selection, anomaly detection, and deep learning leads to a significant boost in emotion recognition accuracy. This research contributes to the field of affective computing and speech analysis, paving the way for more intelligent and emotionally aware AI systems. Emotion recognition through speech processing is an advanced research area that aims to identify human emotions by analysing speech signals. This technology has significant applications in humancomputer interaction, mental health monitoring, virtual assistants, customer service, and psychological assessments. The proposed system employs machine learning algorithms, including Support Vector Machines (SVM), Random Forest, kNearest Neighbours (KNN), and deep learning models like Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs). It utilizes key speech features such as Melfrequency cepstral coefficients (MFCCs), prosody features, and spectral characteristics to accurately classify emotions. This research provides a foundation for creating intelligent and emotionally aware systems that enhance human-computer interaction.