This Letter presents a single-channel speech enhancement technique based on modulation domain signal processing. In the proposed approach, noise-dominant low-frequency (4-16 Hz) temporal modulation envelopes are discarded based on the modulation domain a priori signal-to-noise ratio criterion. Moreover, the authors have employed coherent harmonic demodulation method which ensures distortion-free filtering of modulators. The experiment results in various acoustic non-stationary noise scenario show that the proposed approach can significantly improve the speech quality and intelligibility while compared to the other existing state-of-the-art baseline techniques.