A computer-assisted language learning software for mobile devices (MAC) is presented, that was aimed to helping speakers acquire speech contrasts not native to their own language. The software is based on the high variability phonetic training (HVPT) technique. An overview of the software is given, followed by results from an efficacy study. Two groups using slightly different variations of the MAC software. One group received software training that was equivalent to an ordinary HVPT training, but delivered on mobile devices. The second group received a version that used an adaptive algorithm to determine which words and speakers that the learner had most difficulty with, and gave them more practice on those. Results showed that both groups showed significant but equivalent improvements. These results also showed that the magnitude of improvement using mobile phones was similar to those obtained using fixed PCs.