Raman spectroscopy has emerged as a promising tool for its noninvasive and nondestructive characterization of local chemical structures. However, spectrally overlapping components prevent the specific identification of hyperfine molecular information of different substances, because of limitations in the spectral resolving power. The challenge is to find a way of preserving scattered photons and retrieving hidden/buried Raman signatures to take full advantage of its chemical specificity. Here, we demonstrate a multichannel acquisition framework based on shift-excitation and slit-modulation, followed by mathematical post-processing, which enables a significant improvement in the spectral specificity of Raman characterization. The present technique, termed shift-excitation blind super-resolution Raman spectroscopy (SEBSR), uses multiple degraded spectra to beat the dispersion-loss trade-off and facilitate high-resolution applications. It overcomes a fundamental problem that has previously plagued high-resolution Raman spectroscopy: fine spectral resolution requires large dispersion, which is accompanied by extreme optical loss. Applicability is demonstrated by the perfect recovery of fine structure of the C-Cl bending mode as well as the clear discrimination of different polymorphs of mannitol. Due to its enhanced discrimination capability, this method offers a feasible route at encouraging a broader range of applications in analytical chemistry, materials and biomedicine.