Graded arrays of resonators have proven to be highly effective tools for localizing elastic waves and amplifying local fields. Previous research in this field has demonstrated that these metamaterials significantly increase the displacement amplitude of the resonating components. This, in turn, leads to a substantial increase in extracted electrical power for piezo-mechanical systems. In our study, we demonstrate that this amplification can be harnessed with specific circuits and piezoelectric patches to improve the signal-to-noise ratio. This improvement is achieved by employing resonators operating in close proximity to their first resonance frequency. Introducing an electrical in series connection between each pair of resonators results in halving the output channel while doubling the generated current. The signals from these channels are subsequently analyzed using a machine learning encoder-decoder model to reconstruct the input signal exciting the resonators. This signal is the target of the sensing. Our primary objective is to develop a wave analyzer that leverages the amplifying effects of resonators as a resonating accelerometer. This approach aims to reduce the power consumption of the required circuitry for signal interpretation and to make a first step towards an autonomous sensing tool.