Study of amphibian vocalizations and voiceprint recognition technology in Wuyishan National Park

In recent years, biodiversity has gained increasing attention as a crucial topic within global environmental and development discussions. To understand the baseline and dynamic changes of biodiversity resources in China, investigation and monitoring efforts are essential. However, traditional methods typically require significant time and human resources are constrained by spatial and temporal conditions. Although infrared camera technology has considerably improved the efficiency of monitoring wild mammals, it is less applicable to ectothermic animals such as amphibians and insects. Since animals like frogs, insects, and birds rely extensively on vocal communication, biodiversity monitoring methods can effectively address the limitations of traditional approaches. Based on years of amphibian diversity surveys conducted in Wuyishan National Park, a vocalization database for amphibians in the area was established. This study analyzed the acoustic characteristics of representative anuran species and explored the application of artificial intelligence technology for species identification through amphibian vocalizations. The results indicate that 37 anuran species have been documented in Wuyishan National Park, with vocalization data collected for 30 species, accounting for 81.1% in total. The accuracy rate for species recognition with voiceprint recognition model exceeded 80% for these species. Passive voiceprint recording equipment supports continuous long-term monitoring and can be deployed in areas challenging for personnel to access. Furthermore, the ResNet50 artificial intelligence model can automatically identify the amphibian voiceprint data, significantly enhancing the efficiency of species identification. To sum up, amphibian diversity monitoring technology based on voiceprint recognition presents an effective solution for intelligent monitoring. In the future, the artificial intelligence recognition technology for voiceprint data will play an increasingly important role in biodiversity monitoring and animal conservation.