arXiv AI recent: FAConformer: Frequency-Aware Convolutional Transformer for Auditory Attention Decoding
Researchers proposed FAConformer, a frequency-aware CNN-Transformer framework for auditory attention decoding (AAD).,FAConformer integrates band-specific encoding and adaptive cross-band...
The proposed FAConformer framework consists of independent CNN-Transformer encoders for band-specific modeling and a frequency-aware attention module for adaptive cross-band interaction.,FAConformer also introduces band-wise auxiliary supervision to prevent under-optimization of weakly contributi...