This paper provides a comparative evaluation study with totally blind participants to evaluate their performance using two braille input methods. BrailleEnter, a gesture based input method, is compared to the Swift Braille keyboard, which requires finding the location of six buttons representing braille dots. Six blind participants assessed the performance of both keyboards on Android platforms. There were significant differences between both braille keyboards in term of typing speed, error rate and usability. We found that BrailleEnter was more accurate than the Swift Braille keyboard. BrailleEnter also received higher ratings for usability based on Lewis questionnaire on input methods. At the same time, this study found that the speed of BrailleEnter was significantly slower than the Swift Braille keyboard since it required users to interact six times with the screen to insert just one character. Participants were able to quickly understand the BrailleEnter and Swift Braille input methods. Given their knowledge of Braille patterns, they were also quick to learn how to type on their touchscreen devices using the keyboards. The study indicates both keyboards strengths and weaknesses, and highlights a set of recommendations for developing a more accessible keyboard for blind people. In doing so, we consider possible design improvements for the future development of accessible keyboards for blind users.