This study aims to investigate whether hand gestures mimicking the lip aperture of non-native vowels can improve learners' production accuracy after audiovisual perceptual phonetic training. Sixty-six Catalan/Spanish bilingual learners of English were randomly assigned to either the No Gesture or Gesture group for training on the challenging English vowels /ae/ and /ʌ/. In the Gesture group, participants saw the instructor perform gestures mimicking the lip aperture of the low vowel /ae/ and mid vowel /ʌ/. Participants in the No Gesture group only saw the instructor produce the speech. Pronunciation performance was evaluated before and after training using paragraph reading, picture naming, and word imitation tasks. Pillai scores were used to measure the acoustic overlap between /ae/ and /ʌ/. The results showed that although both training groups showed less overlap between the two vowels after training, gestural training had a greater effect than no gesture training in the picture naming and paragraph reading tasks. These findings suggest that hand gestures mimicking visible articulatory features, such as lip aperture, can improve the pronunciation accuracy of non-native sounds in L2 learners.