Speech technology has become widespread in the past decades: voice assistants, voice biometrics, automated call centres, autotune, and vocaloid idols are part of our everyday lives. These technologies are beneficial and convenient for the general population, but for certain parts of the population, e.g., atypical and pathological speakers, these technologies do not work well. It is especially problematic that pathological speakers are less able to use these technologies as they are often physically disabled, meaning they would have a strong need for voice assistants. Apart from voice assistants, there are several other kinds of speech technologies where the main user would be a pathological speaker. However, these speech technologies currently do not work well for these pathological speakers. In other words, speech technology lacks accessibility to pathological speakers. This thesis presents a series of studies on making speech technology accessible to pathological speakers.