The target of this work was the detection of fault in engine through its sound using spectral analysis. The sound signals of three selected engine parts were examined using spectral analysis to detect the fault in Toyota 3B engine. These sound signals are sound from normal Toyota 3B engine, sound from faulty advanced engine and sound from faulty retarded engine in Toyota 3B engine. These all sounds have a unique spectral representation and have a unique sound which different from the normal engine sound. Sound capturing processes was occurred by combining laptop, mini-jack, microphone, microphone stand and engine and these sound signals were recorded in isolated room to avoid interference from surrounding sounds. Two signal visualization techniques were used in this work and these sound signal visualization techniques are time wave form and line spectrum. The plot of line spectrum was obtained after time domain is converted to frequency domain with help of FFT. This conversion is performed with aid of MATLAB soft ware. This thesis results indicates that each sound signal has its own time waveform, line frequency spectrum and power spectrum. It can be concluded that every sound signal has a unique frequency. This work initiates others to detect the fault in engine using spectral analysis.