The conversion of wind energy into electricity using wind turbine has been widely implemented. However, maintenance is needed so that the wind turbine can function optimally. One of the components which often break is gears. Gears are used to transmit big power and rotation. The gears have teeth around the edges, so that they can transmit power through contact mechanism between the teeth on the gear and the teeth on the pinion. The occurrence of contact load between teeth can cause gears damage on the wind turbine. Therefore, early detection is needed by using vibration signal detection method which aims to analyze the characteristic of time domain vibration signal and the gears frequency spectrum on the wind turbine is normal and one of the gears is broken. This research method is carried out on the gears of horizontal axis wind turbine. The vibration signal acquisition data is implemented on 25600 Hz sampling rate and the recording duration of each data is 5 seconds with 3 seconds pause for each data taking with the axis rotation speed of 1200 rpm. The vibration signal detection method uses frequency spectrum so that gears damage can be detected by counting the gearmesh frequency (GMF) value which appears on each harmonic. However, the amount of noise that appears makes it hard to see the amplitude value of each frequency clearly so that it is hard to detect the gears damage on the wind turbine. Therefore, TSA method is needed to reduce noise which appears from the components which are not being tested. The result from the frequency analysis spectrum before TSA method is used shows that the gears damage cannot be clearly seen because the vibration signal is still full of noise that it makes the amplitude value invisible. TSA method is used to reduce the noise to filter the vibration signal so that it is known that the gearmesh frequency value seen on 1xGMF is 365,027 Hz, 2xGMF is 730,054 Hz