Deep-learning-based framework for inverse design of a defective phononic crystal for narrowband filtering

•This is the first study to employ deep-learning techniques in designing defective PnCs.•A deep-learning-based framework is newly proposed to design a narrow bandpass filter.•Four deep learning models are compared to tackle the nonunique response-to-design problems.•The proposed framework provides an optimal design for the arbitrary defect-band frequency.•The proposed framework facilitates a streamlined and automated inverse design process.