Attention-based LSTM with residual connections for speech emotion recognition

Arxiv CS.SD (Sound/Audio ML)

The paper proposes an attention-based LSTM network with residual connections to reduce computational and memory requirements in speech emotion recognition. The approach aims to maintain high accuracy while lowering model complexity compared to large pretrained models.

speech-emotion-recognitionlstm

Arxiv CS.SD (Sound/Audio ML)·Daniil KrasnoproshinMaxim Vashkevich

8 days ago