Local microphone field calibration can be performed using seismic signals recorded by a well calibrated seismometer. When the 29 May 2004 regional earthquake (Mw 5.1) occurred off the coast of Korea, the Pn, Pg, and Lg seismic phases produced high-frequency (1–4 Hz) local infrasound signals at the Chulwon Seismo-Acoustic Array (CHNAR), an array deployed in the Republic of Korea. Instrument-corrected waveforms of seismic signals are compared to those of local infrasound signals using time-varying coherence estimates to identify the time window and bandwidth for the calibration process. Based on this assessment, the first 6.4 sec of the Lg phase and the coupled infrasound signals from 1 to 4 Hz are used. Envelope functions for the instrument-corrected Lg phase and local infrasound signal are calculated and integrated in the time domain. The transfer function from ground velocity to atmospheric pressure perturbation is estimated in the time domain using the ratio of the integrated envelope functions. The observed transfer functions estimated in this way are compared to the theoretical transfer function based on a model relating ground velocity to pressure perturbation, which is dependent on the density of air and speed of sound at the surface. A maximum difference of 4 dB is apparent between observed transfer functions and theory for highly correlated seismic and coupled infrasound signals (>0.8). Six factors are considered as contributors to these differences, including the porous hose array attached to each acoustic sensor for noise reduction, variations in the density of air or speed of sound, local perturbations in air pressure, differences in ground velocity at each sensor, and differences in acoustic instrument sensitivity. Variations of the acoustic gauge’s sensitivities at CHNAR are identified using a laboratory calibration procedure and documented as the major contributor to the difference between observed and theoretical transfer function at CHNAR.