Computer game playing has been suggested to be an effective training to enhance perceptual and cognitive abilities. Focusing on potential improvements in auditory selective spatial attention induced by computer gaming, we compared a passive waiting-control group with two gaming groups, playing either a first-person audio-only action game requiring spatial attention and sound localization or a platform side-scroller video game without audiospatial components, which has been shown to improve cognitive performance in previous studies. Prior to and immediately after game training for 1 month for at least 30 min per day (total training time ≥15 h), healthy young adults were tested in an audiospatial task simulating a “cocktail-party” situation with multiple speakers at different positions. The proportion of correct target localizations was significantly increased after audio and video gaming compared with the control group. However, there were no significant differences between gaming groups, with similarly strong effects of action audio game and non-action video game trainings on auditory selective spatial attention. Thus, it seems as if successful training of “cocktail-party” listening can be induced not only by modality-specific near-transfer learning within the audiospatial domain, but also by far transfer of trained cognitive skills across sensory modalities, which may enhance domain-general processes supporting selective attention.