Tailored for vehicle horn: A novel sound source capture method

Abstract

The development of a horn sound capture system, capable of replacing manual control, offers a promising solution for alleviating horn noise. Existing horn sound systems lack localization algorithms that are capable of robustly resisting interference. To address this issue, this paper introduces an innovative horn sound capture system, including hardware and algorithms for sound recognition and localization. The pivotal innovation of the system is the introduction of the Sub-SRP algorithm, an advancement of the steered-response power phase transform (SRP-PHAT). Leveraging the frequency characteristics of horn sounds, it employs a series of filtering and spectral subtraction techniques to effectively filter out non-horn noise sources. In the interference experiment and field test, the Sub-SRP algorithm achieved DOA scores of 0.96 and 0.95, respectively, showing an approximately 10% improvement compared to the existing weighted SRP-PHAT, this improvement is even more pronounced under strong interference, increasing to 17%. This improvement significantly enhances the system’s ability to accurately capture honking vehicles and resist interference. Additionally, the horn sound recognition algorithm achieved accuracy rates of 95.75% and 97.65% in the two experiments, demonstrating the system’s effectiveness. In summary, this study contributes to the advancement of urban noise management and intelligent noise control systems.