This study investigates the speaker-discriminatory potential of the Standard Chinese compound vowels in their formant dynamics. The first four formants were extracted from spontaneous speech materials. Two different methods (general polynomials and Legendre polynomials) of fitting formant trajectories of the 8 compound vowels from 85 male speakers were used to compute coefficients as input to a random forest classifier. Besides, original values (averages and raw data) of these formant frequencies were also input to the above classifier as benchmarks. The results show that high-frequency formants (F3 and F4) display a greater discriminatory power compared to low-frequency formants (F1 and F2) in [ai], [ia], [iou], and [uei] while low formants display greater power in the other 4 nasal compound vowels. As for curve fitting methods, Legendre coefficients perform better than the general coefficients in the random forest classifier, while the means perform worst. Furthermore, formants from [ia], [in] and [iou] are better at distinguishing speakers than other 5 compound vowels.