ML & DataPersonal project2024Completed
Speech Sentiment Analysis
Record speech and get a transcript, its sentiment and audio features.

Overview
Wav2Vec2 transcribes the audio, a RoBERTa or DistilBERT classifier labels the sentiment, and librosa extracts pitch, tempo and spectral features. Everything runs in a Gradio app that records audio straight from the browser.