A cross air-water interface hydroacoustic signal detection method based on micro vibration detection at the air-water interface is proposed. Laser speckles modulated by water surface acoustic wave are recorded and used as information carrier. Phase correlation and multi-channel fusion algorithms are used to extract and enhance hydroacoustic signals. Wide frequency range (200 Hz-20 kHz) underwater acoustic signals are detected with frequency relative error smaller than 0.5%. Several common artificial and natural hy-droacoustic signals are used as source signals and correctly reconstructed. The average intelligibility of recovered humpback whale signals evaluated with Normalized Subband Envelope Correlation algorithm is 0.52±0.02.