Urban noise pollution has become an increasingly significant environmental issue, with important implications for human health and overall well-being. Because of this, the need for noise monitoring has become ever more important. In this paper, a low-cost, always-on noise monitoring system is developed to measure sound pressure levels, their corresponding statistical data, as well as their possible origins. The device is based on a Heltec LoRa V4 and an INMP441 MEMS microphone using LoRaWAN communication architecture. A neural network is implemented via Edge Impulse, allowing noise classification, which is an important feature for noise guideline compliance and for mitigation strategy implementation. Measurements and noise classification are sent to a GitHub repository, which is then displayed in a Streamlit-based app. The device provides accurate measurement of noise levels on a 24-hour basis, allowing for better implementation of mitigation strategies. The resulting device correctly measures sound levels within a ±2 dB margin of error when compared to the measuring reference.