This study explored the potential of adopting speech-to-text recognition (STR) technology for English as a foreign language (EFL) oral training in class at the university level. An action study method was set to investigate the effects of implementing STR app tasks for EFL oral training with 27 students in one class for one semester at a university located in Taiwan. Data were obtained through pre- and post-tests, speaking practice results, field-observation notes, student reflective journals, and end of class survey. The results of the quantitative data analysis indicated that the STR app tasks were effective in increasing students' English speaking ability. Students also expressed positive attitudes toward the use of the tasks in the STR app. Further, the qualitative data analysis showed that the students found these tasks highly motivating and can quickly engage them in learning. The STR app tasks that provide repetitive training of spoken English, particularly on pronunciation, fluency, and vocabulary acquisition approved be more beneficial to learners than traditional teaching methods. It is highly recommended for teachers to design a variety of STR tasks to meet individual learners' needs and preferences.