Speech Emotion Classification with novel Parallel CNN-Transformer model built with PyTorch, plus thorough explanations of CNNs, Transformers, and everything in between

This repository contains PyTorch implementation of 4 different models for classification of emotions of the speech.

An ML-powered app that classifies emotions in speech (happy, sad, angry, etc.) using the RAVDESS dataset. Audio features are extracted with librosa, and emotion classification is performed using scikit-learn or PyTorch models. Ideal for emotion-aware applications in voice interfaces.

No description available

Emotion classification from speech using convolutional neural networks, mel-spectrograms, and data augmentation with PyTorch.

A multi-task deep learning system for Speech Emotion Recognition and acoustic phenotyping. Features dual-branch wav2vec2 + handcrafted feature architecture, Supervised Contrastive pretraining, multi-head classification, and clinical inference pipeline. Built with PyTorch on RAVDESS, CREMA-D, and EMU corpora.