This research aimed to find out the improvement of students’ pronunciation in word stress through text-to-speech software at the 11th-grade students of SMAN 1 Kaledupa, in the academic year of 2021/2022. The samples of this study were 22 students, and purposive sampling was applied for this research. This research used a pre-experimental research design where there were three sessions conducted to reach the result of this research. Pre-test and post-test design consisted of 5 meetings, including the treatments. The pre-test in this research is to obtain students' prior knowledge, and the post-test is to measure students’ improvement after the researcher applies or uses the text-to-speech software in the treatment session. The finding of this research showed a significant improvement in students’ pronunciation after the use of text-to-speech software. This was proven by the mean score before treatment was 32.04, and the mean score after treatment has given was 60. The percentage of the improvement was 87.26%. The t-test value (6.106178) was higher than the t-table value (2.07961). H1 is accepted while H0 is rejected, and it can be concluded that there is a significant improvement in students’ pronunciation in word stress.