It’s rare that I can hear a recorded voice and be unable to distinguish it from the live version. So rare, in fact, that I think the only times I have been fooled is when that voice is distantly recorded and I am not focused on it.
I place the blame on the device that captures those voices, the microphone. Recording technology has gotten good enough that a direct injected signal from an instrument is indistinguishable from its live performance. It’s really only microphones that are so antiquated as to be instantly identifiable. With only a small amount of practice, you’ll find it easy to classify the types of colorations microphones add to the human voice.
It’s curious to me that so little innovation has gone into the development of new microphones. But then, I suppose, it’s a very difficult problem. To capture soundwaves you pretty much have to move some amount of mass and then measure that movement. The types of mass being moved: plastic, metal, ceramic, or granular materials contributes to sound’s colorations.
I once daydreamed about building a servo system around a microphone where a difference approach could be used to correct for the colorations imparted by various materials and microphone patterns. It was an interesting idea but quickly abandoned because, of course, there was no way of capturing what it should sound like without using another transducer.
No, the problem of microphones will be with us for a long, long time.