Speech-based technology is a useful alternative to traditional input techniques such as the keyboard and mouse. For people with disabilities that hinder use of traditional input devices, a hands-free speechbased interaction solution is highly desirable. Various speech-based navigation techniques have been discussed in the literature and employed in commercial software applications. Among them, grid-based navigation has shown both potential and limitations. Grid-based solutions allow users to position the cursor using recursive grids to 'drill down' until the cursor is in the desired location. We report the results of an empirical study that assessed the efficacy of two enhancements to the grid-based navigation technique: magnification and fine-tuning. Both mechanisms were designed to facilitate the process of selecting small targets. The results suggest that both the magnification and the fine-tuning capabilities significantly improved the participants' performance when selecting small targets and that fine-tuning also has benefits when selecting larger targets. Participants preferred the solution that provided both enhancements.