@Cyber sleuth
First, I return your kind words and wish you and your loved ones a happy Easter Sunday, as well as to all of you reading along.
I wanted to comment on the audio comparison you are speaking about. The analysis a court does when determining whether a voice identification is allowed under the rules of evidence comes from common law, in particular the case of
Neil v. Biggers, 409 U.S. 188 (1972). I have included a link below for you to read should you be interested.
Link:
Neil v. Biggers, 409 U.S. 188 (1972)
This case dealt with both eyewitness and voice id, and a rule was formed by the Court to test reliability of these identifications. The test requires courts to consider two things: (1) the suggestive nature of the identification; (2) the reliability of the identification. As far as suggestive, think one on one id's vs. photo lineups. As far as reliability, the court laid out "indicia" of reliability in
Manson v. Brathwaite, 432 U.S. 98 (1977).
Link:
Manson v. Brathwaite, 432 U.S. 98 (1977)
Without being more verbose than I have already been, I will simply say the way you have described recognition would not pass muster on a due process level. A one-on-one comparison between RA and BG voice samples would first be overly suggestive. Further, such a comparison would also be unreliable, as layperson identifications (i.e. the jury) are notoriously unreliable and inaccurate. The jurors will have never heard RA's or BG's voice before, and are thus not sufficiently familiar with it to make such an ID. There were also no known witnesses who heard RA's or BG's voice, so hearsay would also be a problem.
Here is a really good law journal article on the topic should you be interested. I think it will help those reading in thinking about not only voice analysis comparisons, but also visual identifications.
Journal:
https://repository.uclawsf.edu/cgi/viewcontent.cgi?article=3507&context=hastings_law_journal