In murder mysteries one of the classic plots is playing a recorded conversation in order to pretend that the victim was alive later than he was. Typically used to allow the murderer to establish an alibi for the supposed time of death, the curious thing about this trick is that as technology makes it easier for the murderer to do, it also makes it harder for the murder mystery writer to use.
Given how many murder mysteries were written shortly after the initial Sherlock Holmes short stories in the early 1890s and are now largely lost to history, making any statements about firsts in a murder mystery is always fraught with peril. That said, the first time that a recorded conversation was used to pretend that a victim was alive later than he was, that I’m aware of, is Agatha Christie’s The Murder of Roger Ackryod. First published in 1926, it used, perforce, a dictaphone which at the time would cut grooves in wax on a cardboard cylinder. The sound quality actually achievable on such a machine would not be great, which makes its believability to listeners somewhat questionable. On the other hand, it was played behind a closed door. The door would muffle the sound, making it harder to tell a recording from a real voice. And then there’s the psychological aspect, which gets to the heart of the modern problem—at the time, people’s voices being recorded was so rare that no one would expect it, so they wouldn’t think to look for the differences. We are used to dealing with imperfect sound and concentrating on the words; without paying careful attention we are likely to ignore the problems we don’t recognize in a recording since we’re not used to hearing them as the characteristic signature of a recording.
These days, high fidelity recording devices are extremely cheap, to the point where they are a standard component of many devices including the cell phones that everyone carries around in his pocket. Decently loud and accurate playback devices are so common that they can be found in novelty birthday cards meant to be thrown away after use. If a killer wanted to use a recording of a victim to create the impression of a later time of death it would be cheap and easy, and the playback would be of such high quality that it would take a trained ear to have a chance of telling whether it was a recording.
Ironically, that’s the problem. Playing the victim’s voice would be so easy that if a reader is presented with someone having merely heard the victim’s voice without actually seeing him, he immediately suspects that it was a recording. Admittedly, the same is true of the low-tech gambit of dressing in the recognizable clothing of the victim and only being seen from a distance, without talking, too. I suspect it’s only a matter of time before we get pre-recorded video calls to establish a fake time of death, and some day (when the prices come down, perhaps) a “deep fake” where powerful video software uses images and mapping techniques to synthesize new video of a subject.
The real trick is to make it seem natural to have something in the way of a witness seeing the victim with his own eyes, up close. The moment there is something in the way, the reader’s suspicions will be (justly) aroused. That’s the trick, but it’s a very difficult one. So difficult, in fact, that I’m not sure how to pull it off.