The on-load tap changer (OLTC) is a key component of the power transformer. The operating state of OLTC directly affects the stability and safe operation of the power system. Therefore, online monitoring of the operating state of the on-load tap changer is quite crucial. This study utilizes the fiber grating acoustic sensor and the motor current sensor to monitor the operating state of OLTC online, extracts the motor working time and the root mean square value of the motor current, and uses short-time Fourier transform to extract the frequency characteristics of the OLTC switching signal. Mechanical faults were identified through the voiceprint features of normal and different fault states, and OLTC spring breakage and contact loosening mechanical fault experiments were carried out to verify the feasibility of the method. The results show that there is no significant change in the operating time and the root mean square value of the current of driving motor due to OLTC switching core faults such as broken springs and loose contacts. When it comes acoustic signals, the spring breakage fault and the contact loosening fault cause high-frequency components of approximately 12 kHz and 50 kHz at the moment of contact impact, which is different from normal operating conditions where most frequency components are below 10 kHz. This study provides new ideas for the research on online monitoring methods of OLTC.