CuBiAAD: A cue-masked bimodal auditory attention dataset based on EEG and fNIRS

Investigating auditory attention in multiple speech streams is crucial for the research on language comprehension and the development of neuro-steered hearing devices. This study proposed a multimodal dataset recording brain activities based on a cue-masked auditory attention paradigm, which avoid information leakage before the experiment for better simulating real-world scenarios. The dataset included electroencephalography (EEG) data from 30 participants, functional near-infrared spectroscopy (fNIRS) data from 10 participants, and all the audio files played during the experiment. To the best of our knowledge, this was the first multimodal dataset based on auditory attention, and it was currently the dataset with the largest number of participants. The dataset would benefit the studies on the mechanism of speech comprehension, the sensor optimization and algorithm design of auditory attention decoding, facilitating the technology development of auditory-based brain computer interface (BCI).