LSTM Speech & Audio Emotion Classifier

Overview

A Deep Learning pipeline that detects and classifies human emotions from speech and audio signals. This project leverages Long Short-Term Memory (LSTM) networks to capture the temporal and sequential dependencies inherent in audio time-series data.

Architecture & Technical Stack

• Feature Extraction: Audio signal processing (e.g., Mel-Frequency Cepstral Coefficients - MFCCs) to extract spectral features from raw audio waves.
• Model: Multi-layer LSTM network designed for sequence classification.
• Domain: Deep Learning, Audio Signal Processing, Affective Computing.

Project Structure

• /data - Raw and preprocessed audio files (ignored in git)
• /notebooks - LSTM_Speech_Audio_Emotion_Detection_and_Classification.ipynb containing EDA, feature extraction, and model training.
• /models - Saved .h5 model weights (ignored in git)