Two experiments are reported upon the influence of a computer aid on the recognition of complex sounds. Twelve112‐htraining and test sessions were conducted over a period of two weeks. Measures of unaided listener performance (L), computer performance (C), and combined performance (C&L) were obtained. In Expt. 1 the reliability of the computer was varied, while audio signal‐to‐noise (S/N) ratio was kept constant. The results showed (1) that as C increases so (C&L) increase; (2) that excessively high C performance reduces the listeners performance; and (3) that the strategy involved in using the computer and audio displays tends to evolve with training. Initially, attempts to compromise between computer and audio information often lead to (C&L) performance lower than it need be if the poorer display had been ignored. Later, the more valid display tends to be relied upon and finally in certain instances an optimal strategy tends to emerge in which (C&L) performance is above both C and L. Experiment 2 compared “dependent” and “independent” systems. In the former condition audio and computer displays are fed to the same observer; in the independent condition one observer receives the audio imputs and a partner views the computer display and makes the final (C&L) decision. No major differences in overall preferences were observed, except that the independent condition yields poor L performance.