Self-delivered speech therapy provides an opportunity for individualised dosage as a complement to the speech-therapy regime in the long-term rehabilitation pathway. Few apps for speech therapy have been subject to clinical trials, especially on a self-delivered platform. In a crossover design study, the Comprehensive Aphasia Test (CAT) and Cookie Theft Picture Description (CTPD) were used to measure untrained improvement in a group of chronic expressive aphasic patients after using a speech therapy app. A pilot study (n = 3) and crossover design (n = 7) comparing the therapy app with a non-language mind-game were conducted. Patients self-selected their training on the app, with a recommended use of 20 minutes per day. There was significant post-therapy improvement on the CAT and CTPD but no significant improvement after the mind-game intervention, suggesting there were language-specific effects following use of the therapy app. Improvements on the CTPD, a functional measurement of speech, suggest that a therapy app can produce practical, important changes in speech. The improvements post-therapy were not due to type of language category trained or amount of training on the app, but an inverse relationship with severity at baseline and post-therapy improvement was shown. This study suggests that self-delivered therapy via an app is beneficial for chronic expressive aphasia.