Amide I vibrational spectra of proteins can provide critical information on secondary structure elements and structural inhomogeneity. Despite that there exist a number of linear and nonlinear vibrational spectroscopic studies reported, it has been quite difficult to quantitatively simulate the amide I vibrational spectra of polypeptides and proteins. To achieve this goal, we have developed theoretical and computational methods such as constrained molecular dynamics simulation, Hessian matrix reconstruction method, and fragmentation approximation to describe delocalized amide I normal modes of proteins as linear combinations of amide I local modes. By using the computational scheme, amide I IR, vibrational circular dichroism, and two-dimensional IR photon echo spectra of polypeptides and protein in solution were simulated and compared with experimental results. The structure-spectrum relationships established are discussed. It is believed that the present computational method will be of use in shedding light on the underlying vibrational dynamics of protein as well as in interpreting experimentally measured linear and nonlinear amide I spectra of proteins in the future.