💻 🤖 A summary on our attempts at using Deep Learning approaches for Emotional Text to Speech 🔈

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

[ICASSP 2023] Official Tensorflow implementation of "Temporal Modeling Matters: A Novel Temporal Emotional Modeling Approach for Speech Emotion Recognition".

Offical implementation of paper "MSAF: Multimodal Split Attention Fusion"

[RAVDESS] Speech Emotion Recognition with Convolutional Attention based Bi-GRU. (Best test accuracy of 87%)

Classifying Audio to Emotion

1,440 audio files (.wav), i.e. speech files, from 24 actors that are categorized into 8 separate emotions.

In this work is proposed a speech emotion recognition model based on the extraction of four different features got from RAVDESS sound files and stacking the resulting matrices in a one-dimensional array by taking the mean values along the time axis. Then this array is fed into a 1-D CNN model as input.

CNN-LSTM based SER model using RAVDESS database

Use GANs to generate spectrogram of speech. The generated spectrogram will be conditioned on emotion

Emotion is an intuitive feeling which can be determined from any person’s circumstances and surroundings. But in this project, we tried to identify the emotional state of a person using his voice as input.

A convolutional neural network trained to classify emotions in singing voices.

Pytorch speech emotion recognition for RAVDESS dataset with CNN.

Collaborated to create a Machine Learning model trained and tested with a Random Forest model to predict primary emotion based on input audio file. Data cleaned and trained in a Jupyter Notebook using Pandas and Librosa. Results visualized using Pandas, Tableau, and JavaScript functions with bootstrap in a dynamic HTML website.

This repository is an import of the original repository that contains some of the models we had tested on the RAVDESS and TESS dataset for our research on Speech Emotion Recognition Models.

Comparing text-only (RoBERTa), audio-only (wav2vec2/WavLM), and multimodal transfer learning approaches for Speech Emotion Recognition on RAVDESS, MELD, and IEMOCAP. PyTorch Lightning + HuggingFace + Gradio demo.

This repository contains the source code for my final year project for my undergraduate degree in MTU.

Speech Emotion Recognition system using CNN architecture to classify 8 emotions from audio recordings (RAVDESS dataset). Features depthwise separable convolutions for efficient audio analysis, reaching ~70% accuracy.

This project explores facial emotion recognition by combining spatial and temporal learning in a structured two-stage pipeline. The motivation was to move beyond static image-based classification and build a system capable of understanding how emotions evolve over time in real-world scenarios.