In this paper, an adaptive method of cancellation of parasitic vibrations is presented for a Self-Mixing (SM) interferometric laser vibration sensor which has been coupled with a solid state accelerometer (SSA). Previously, this was achieved by using a pre-calibration of phase and gain mismatches over the complete bandwidth of the instrument. Such a pre-calibration is not only tedious to execute but also hinders a mass production of the instrument as every SSA-SM sensor couple requires customized calibration. On the other hand, the proposed method does not require any pre-calibration as it uses an adaptive filter that self-tunes to match any unknown phase and gain differences between the SSA and the SM sensor. Two different adaptive algorithms, namely Recursive Least Squares (RLS) and Least Mean Squares (LMS) algorithms are tested and a comparison is established on the basis of parameter dependence, convergence time, computational cost and rms error. The proposed algorithms have provided improved results (mean error of 19.1nm and 20.2nm for LMS and RLS respectively) as compared to pre-calibration based results (mean error of 24.7 nm) for a laser wavelength of 785 nm. Simulated and experimental results thus demonstrate the utility of such an approach for embedded vibration sensing corrupted by extraneous parasitic motion. Index Terms-adaptive filter, self-mixing, vibration measurement, embedded sensing, optical feedback, laser instrument.