Wall Street English has built online activities that allow students to record phrases and receive word-level Automatic Speech Recognition (ASR) driven pronunciation feedback. Students in language centres in China, Vietnam, Saudi Arabia, and Italy (N=2,867) used ASR-Computer Assisted Language Learning (CALL) activities, and some (N=482) completed a questionnaire. A high number of students reported that ASR-CALL activities helped them to improve their pronunciation. However, the study found remarkable differences in usage of product features across countries, with students from Vietnam and China using more retries than Saudi Arabia, and students from Italy using the fewest retries. Students from China, Vietnam, and Saudi Arabia more frequently listened to model audios than students from Italy. A series of Kruskal-Wallis tests revealed significant group differences between dominant L1 and students’ beliefs and perceptions using ASR, and between age groups and students’ beliefs and perceptions using ASR. This study points to the importance of considering regional differences, and suggests that learner engagement may depend not only on the effectiveness of the technology, but also on learner beliefs and perceptions.