The signals processing algorithm is presented based on laser spectroscopy direct absorption signal (DAS) and wavelength modulation spectroscopy (WMS) for the trace carbon monoxide (CO) measurement. The simulated transmittance data of the pure CO gas are from the HATRAN database. The DAS intensity, appending WMS 1-f and WMS 2-f signal intensities are used as the raw signals. Aimed to obtain optimized filtering algorithm, those raw signals which were added Gaussian white noise are denoised by using diverse wavelet-bases and decomposition layers. The effectiveness is validated by our CO concentration detection experiment which measures the weak laser absorption spectral line P(4) of second overtone band at 1.578 μm. A 0.95 m Herriott-type cell provides an effective absorption path length of 55.1 m. Comparing the sensing performances without and with using the optimized wavelet, the experimental results show that the signal-to-noise ratios of the system are significantly improved by 1 to 2 orders of magnitude for the DAS, 1-f and 2-f signal. The anti-jamming capability of the system is improved by proposing the suitable wavelet-base and decomposition layer algorithm.