For the audio that you can hear, a WAV file, which is uncompressed, will catch everything that you could hear on the tape in the range of what a human can hear over maybe a dynamic range of maybe 60dB (my guess). The 20-20KHz frequency range is somewhat of a crap shoot as the upper end is often -10 to -30 dB down and the response curve is anything but flat. The typical 44 or 48 KHz WAV file digitization is far better than the cassette tape. Just make sure the recording is not clipping on peaks, compressed in any way (no MP3), not noise reduced (Dolby), and not "enhanced" by the WAV recording software. While these may make the audible part of the recording easier on the ears, it also causes low level artifacts to disappear forever. Such "enhancements" can always be done on the WAV file later. I'm not sure if a FLAC file is appropriate.
Conspiracy theories are not going to save the earth. More likely, the information is going to be part of some media circus, of which I want no part.
Yeah, sorta. The upper limit on a cassette recording is about 20 KHz. Digitizing much over the Shannon limit (2*20KHz = 40 KHz) isn't going to magically deliver sounds on the tape that were over about 20 KHz. However, what a higher frequency will do is prevent the 40 KHz digitization rate from creating aliasing frequencies that will appear below 20 KHz and mangle the digitized copy. 96 KHz 24 bit and 192 KHz
24 bit digitizers are cheap and common. Use one.The problem for this recording is NOT the electronic copy, which is potentially of far better quality than the original cassette tape. The problem is that the cassette tape is by its very nature a piece of disgusting technology that should have died before it was inflicted on the general public. It would be difficult to design something of lesser quality. Even 8 track was better. When you digitize the audio (hopefully not using a microphone like the CNN reporter), it will probably not be worse than the original cassette, but also not any better. In other words a close to perfect reproduction.
The next step would probably be to "clean up" or "enhance" the digitized WAV file recording. Over simplified, you can emphasize any part of the recording, at the expense of other parts of the recording, in either the frequency domain, time domain, or amplitude range. However, there's no free lunch. To make low level noises more audible, you have to limit, clip, chop, compress, or otherwise reduce the high level sounds. To bring out sounds with a limited frequency range, you have to reduce the level of other sounds outside this range. To bring up the level of gunshots and echos, you'll need to reduce other loud noises. Anyway, welcome to forensic acoustic analysis.