Abstract With the continuous development and improvement of China’s power system, the length and coverage of transmission lines are getting larger and larger, making it inevitable that transmission lines will pass through areas with harsh environments and complex geographical environments. If defects are not discovered and handled properly, over time, serious losses will be caused. Traditional manual inspections can no longer meet actual requirements due to low efficiency. This paper proposes a transmission line defect identification method based on voiceprint recognition. First, the sound signal collected by the sensor is divided into frames and windowed for preprocessing. Then, feature extraction is performed based on the Mel frequency cepstral coefficient. Finally, the extracted vector is used to train the nearest neighbor classifier, and the trained classifier can identify defects in transmission lines.