Summary of a research report by Paul T. Brady that suggests that absolute pitch ability can be learned and improved through practice (note that "perfect pitch" is usually called by researchers "absolute pitch"):
Paul T. Brady
"With intensive training, the author, who had no previous ability in identification of randomly presented tones form the 12-tone musical scale, achieved a performance level of 65% exactly correct and 97% correct within +/- half-step (semitone). The training task consisted of spending many hours hearing and identifying randomly presented tones with feedback on the judgements. Successful acquisition of absolute-tone identification contributes to the evidence that an adult can learn this task . . . " "The breakthrough occurred in May when I began to apply a version of the technique suggested by Cuddy (1968). It consists of playing sine waves at frequencies randomly selected from the musical scale, at first heavily weighted by C's (C is arbitrary; any note will do), but with the proportion of C's eventually dropping to 1/12. The non-C's were always uniformly distributed. Cuddy required her subjects to identify only the anchor tone, but I tried to identify all tones. . . . At first the tones (1 sec long) occurred about every 5 sec, with high probability of C (0.4). Eventually, they occurred with uniform probability every 1.4 sec."
"Fixed-Scale Mechanism of Absolute Pitch," by Paul T. Brady, in The Journal of the Acoustical Society of America, 1970, Vol. 40, 883-887.
Notes on Paul T. Brady's article: The practice method Brady uses (suggested by Lola L. Cuddy) is very interesting, and can be implemented in EarTest very easily. Simply change the "weight" of the note C on the EarTest main screen to be much higher than the other notes. Then over a period of weeks or months, gradually reduce the weight of C until it is equally weighted with the other notes.
Brady suggests the absolute pitch can be learned "only with great effort". He listened to his training tapes for about 60 hours before acquiring an apparently long-term and stable absolute pitch ability. Although this many hours is a lot of work, it is worth pointing out that 60 hours of listening and work is about what you would spend in one semester of a university or conservatory ear-training course. My first university ear-training class met 5 hours per week for 15 weeks--a total of 75 hours. In addition to the class time, we had outside of class assignments. In that time, we learned a lot, but we didn't learn perfect pitch! With this college class as a comparison, spending 60 hours to develop a solid, permanent sense of absolute pitch seems quite reasonable--a good investment of time for a musician.
When Brady tested himself for absolute pitch ability, he had someone play him one note per day, first thing in the morning, for 57 straight days. So there is little doubt that this is real absolute pitch ability, not simply a highly developed sense of relative pitch that happened to get lucky on a single test on a single day.
EarTest is a Windows program that allows you to practice perfect pitch (and relative pitch) ear training. EarTest is flexible--you can try all the practice methods suggested by these studies.