emotion2vec is a speech emotion recognition model open-sourced by Alibaba DAMO Academy, trained on a self-supervised learning framework, capable of extracting emotion embedding vectors from raw audio waveforms.