This study investigated the potential of phonetic symbols and keywords as response labels for perceptual training of L2 sounds. Seventy-one Spanish learners of English were assigned to three groups: symbols, keywords, and control. Students in the symbols and keywords groups followed a 4-week High Variability Phonetic Training (HVPT) program based on identification tasks. The target aspects addressed were eight English vowels that tend to be problematic for Spanish EFL learners (/i: I ae [lambda] 3: e [backwards a] [backwards c]:/). Training stimuli consisted of consonant-vowel-consonant (CVC) nonwords featuring these vowels on a range of phonetic contexts. Overall, the results revealed significant differences between the perception gains made by the two experimental groups, which performed similarly, and the control group. Both experimental groups were able to transfer gains to untrained nonwords, and to untrained voices. Moreover, gains were maintained over time. Improvements were also made in real words, especially by the symbols group, although no significant differences were found between groups. The results suggest that both phonetic symbols and keywords are effective labels for perceptual training and the creation/consolidation of perceptual sound categories. The study offers further evidence of the effectiveness of HVPT for pronunciation training as well as implications for perceptual training studies and language teaching.