Bird Sounds Recognition Algorithm Using a Time Delay Neural Network

Bird monitoring is an effective method to investigate and research the environmental health status. Traditional bird monitoring relies on labor-intensive manual methods while computer vision monitoring is constrained by the environment and is not suitable for all bird monitoring scenarios. This paper proposes a time delay neural network-based bird sounds recognition algorithm that builds a dataset, extracts audio features and uses an enhanced time delay neural network for training to achieve efficient bird sound recognition. The algorithm was then tested on the collected audio data and proved to be effective.