A system to assist speakers with physical disabilities by transforming their speech into a form easily understood by listeners. This technology takes speech signals spoken into a microphone and corrects various degradations caused by problems in speech production. The system corrects the tempo of speech, the frequencies of vowels, can remove acoustic noise, and even replace dropped sounds (e.g., producing the word "not" instead of "naw") given the lexical context. The result is speech that can be up to twice as understandable to the typical listener. This system is highly extensible and can be embedded, for example, within a mobile device.