We propose enhancements to score-based generative modeling techniques for low-latency pilot-based channel estimation in a point-to-point single-carrier multiple-input multiple-output (MIMO) wireless system. Building on recent advances in score-based models, we investigate a specific noise schedule design and sampling acceleration by step-skipping to reduce the number of denoising steps during inference. We additionally propose a single-step signal-to-noise ratio informed denoiser as an extreme case of the step-skipping approach. Our methods achieve significant latency reductions without performance degradation, as demonstrated on a synthetic channel dataset representing an urban macrocell MIMO communications scenario.