This study analyzes traffic signal cycle duration to harmonize traffic flow and minimize noise pollution on urban roads. Data were gathered from nine major intersections in a tier 2 city during the morning (7:30–10:30 AM) and evening (5:00–10:00 PM) time, at half-hour intervals. The collected dataset includes traffic variables such as signal durations, vehicle types and numbers, queue lengths, PCU, and associated noise levels (Leq). Correlation analysis revealed noise levels increased with queue length and decreased with longer green signal durations. For the current analysis, noise levels were estimated using XGboost regression model for varying green signal durations. The analysis demonstrated that setting the green signal duration to 75 s effectively balances noise reduction and permissible traffic flow rates, resulting in a projected noise level of approximately 64.45 dB. These findings highlight the critical role of signal timing and traffic composition in mitigating noise pollution. Leveraging machine learning to model intricate traffic behaviors, the study underscores that optimizing signal timing can substantially enhance the quality of life for urban residents.