Pronunciation has been an important topic in the field of English education. However, students’ pronunciation performances are often difficult to be evaluated statistically and systematically with instant feedback. To solve this problem, some studies applied Automatic Speech Recognition (ASR) system to help students correct the pronunciations. Yen (2019) used Learn Mode to help Taiwanese senior high school students to solve their English pronunciation problem, however, very few elementary school students in Taiwan have used this system to help with their pronunciation problems. This study aims to demonstrate how the phonetic-based ASR system, Speech Ace, could provide elementary school students phonetic-based feedback and identify the problems more precisely by investigating the effect of applying Speech Ace to improve young EFL students’ pronunciation performance. This study provides the analysis of 130 fifth grade elementary school students’ pronunciation performances and the relevance between ASR and their performances. The participants received pronunciation instructions focusing on 3 topics, including “m, n and ŋ”, “ɔ and ɑ” and “ɛ and æ” as a reviewing activity. After the instruction, the participants use Speech Ace to practice with the target words. The results are recorded and analyzed to compare the difference between the participants’ pronunciation performances before and after using Speech Ace. Our preliminary findings show the participants’ pronunciation performances improved through online practices with the ASR system, Speech Ace. The students are more engaging to improve pronunciations with gamification and multimodality based on the system, initiating positive peer competition and self-learning motivation. With ASR system, teachers can help students clarify the results more efficiently and guide the students how to practice more specifically to improve their pronunciations.