One of the major goals of mining science is to develop models that connect monitoring signals with the dynamics accompanying hazards. Whether the triggering process of an outburst produces observable signatures is a fundamental question related to the potential for outburst early warning realization and probabilistic forecasting. Inspired by a recent discovery that certain volcanoes often generate infrasound waves before eruption, we hypothesize that a similar infrasound-generating mechanism occurs in select outburst-triggering processes. In the laboratory, we designed an experimental program to verify our hypothesis, which considered the presence of a gas pocket at the trigger stage. The results indicated that periodic infrasonic tremors spontaneously emerge if gas temporarily accumulates in the gas pocket before escaping from the simulated coal seam. Moreover, the tremor frequency systematically decreases with increasing pocket thickness. Therefore, we proposed a viable model to explain the origin of these periodic tremors. In particular, our model demonstrated that the pressure in the gas pocket is governed by the equation of a linear oscillator. Hence, under proper conditions, infrasonic tremors emerge due to periodic pressure oscillations in the gas pocket. More importantly, our model predicted that different outburst-triggering processes (e.g., fracture development and excavation) can leave distinct footprints in the tremor properties. If this model holds true for actual coal seams, it could be possible to predict outbursts. Our tremor model also opens new perspectives for dynamics monitoring within coal seams.