Abstract: Abstract: Chute blockages, as a common issue in industrial production, not only affect production efficiency but also pose safety risks. Thus, real-time and accurate detection of chute blockage is of significant importance. However, traditional detection methods encounter various problems and challenges in practical applications, such as low accuracy and dependence on manual intervention. This paper proposes an innovative solution based on audio information to address these challenges. We construct a deep neural network model called WaveGNet, which combines WaveNet and GRU, for chute blockage detection. The model extracts and analyzes sound signals to identify characteristic patterns associated with blockages, achieving accurate detection. WaveNet excels in extracting high-quality sound features, while the GRU network captures temporal relationships within sound sequences. The fusion of these two components enhances our understanding of sound signals, enabling more accurate analysis in both the time and frequency domains. This approach reveals patterns relevant to blockage states, thereby enhancing detection accuracy and robustness.Our research directly captures blockage states through sound information, reducing the need for manual intervention and ensuring real-time capability. Our method offers an innovative, efficient, and reliable solution for chute blockage detection in industrial production, holding great potential in practical applications.