This pilot study explored the use of AI-generated speech to support short-term memory and improving skills in consecutive interpreting (CI) training for English majors at the Diplomatic Academy of Vietnam (DAV). CI demands strong memory and quick mental processing, making it a challenging task for learners. Eight third-year students were divided into two groups: one trained with AI-generated speeches, and the other with traditional human-voiced recordings. Over four weeks, both groups practiced using materials that gradually increased in difficulty by adjusting speech speed, content density, and complexity. The preliminary results proposed that the students using AI speech made greater progress, especially in tasks requiring higher cognitive effort. They also reported that the structured and gradually challenging nature of AI practice helped them feel more confident and focused, despite occasional technical issues with audio quality. Overall, the findings suggest AI-generated speech can be a valuable tool in CI training. It offers consistent and scalable practice that effectively targets key memory functions. However, the students still struggled with note-taking and vocabulary, highlighting the need for ongoing instructor guidance. A combined approach using AI tools along with teacher support is recommended to strengthen interpreting skills.