I know from experience that hearing a recording of my voice sounds odd to me, and it doesn’t matter if it’s a digital or analog recording. The reason, I believe, is because how I hear, e.g. using my two ears, is very different from what a microphone in front of my face picks up. But would a high quality recording of my voice sound exactly the same as my live voice to someone else? In other words is there something inherently different in a recorded voice that distinguishes it from a live voice, or could someone be easily fooled?
The way you hear your own recorded voice is how other people hear your live voice.
Our body transmits vibrations internally. The microphone, and other people, are picking up the vibrations transferred through the air. That’s an inherent difference.
Upon careful reading of the question, the issue of hearing one’s own voice is a red herring. The question is about the fidelity of a recorded speaking voice.
The audio recording industry has sought for decades to accurately reproduce live sound through high-fidelity recording and playback technology. There is no magic secret ingredient in a live voice that is lost when recording it. But theoretically a high-fidelity playback of a high-fidelity recording of a speaking voice could be indistinguishable from a live person, although in practice I am sure that most people could tell the difference given today’s audio technology.
This is because a live person speaking in a room is a complicated audio source. A speaker is effectively a point source, whereas the sound is generated by vocal chords, and the sound is conducted through a complex system of resonance chambers (i.e., chest, throat, sinuses). It is somewhat directional. All of this affects how we hear the voice with two ears and a brain due to tone, phase changes, how sound is reflected on room surfaces, and so forth. Also, audio systems are not optimized for recording and playing back a single human voice. I suppose that if a system were designed specifically for that, it could fool you. But they are optimized for music.
On the other hand, it is impossible to distinguish between a live voice and a recorded one when the live one reaches you through a speaker, like hearing the singer at a concert or talking on the phone.
Right,I get that, but when you are hearing something do you hear it using something other than your ears? So a recording of my voice versus my actual voice sounds exactly the same to someone else, correct?
So if I have a good enough microphone, and a good enough recording system, I could record my voice and play it back to someone blindfolded and they couldn’t tell the difference between the recording of my voice and my actual speaking voice. Correct?
I missed your post CookingWithGas. That makes sense to me.
I can recognize hundreds of voices, including bare acquaintances, live or recorded on my answering machine. The one that always astonishes me is my own.
wrong
The poor dynamic range of the magnetic media gives it away…
That’s peculiar, because my recorded voice sounds much deeper to me than the voice I hear when I speak.
The main reason your recorded voice sounds different from your live voice to you is that you are hearing your live voice transmitted both through the air and through your skull bones. In this sense your recorded voice will sound a lot closer to your live voice to other people than it does to you. But, it won’t sound exactly the same because when you record a voice you are also recording the reverberation effects of the room. When you play it back you get the combined reverb of the recording room and the playback room. Your live voice only ever has the acoustic effects of the room you are in. You can minimise this with how you record your voice but it will never be eliminated completely. Can you minimise it to the point it’s inaudible? I don’t know.
It sucks, doesn’t it?
Somehow the rich masculine baritone voice I always hear in my head just doesn’t reproduce well on … anything. Clearly it’s the fault of the equipment, since I would hate to find out that that’s how my voice really sounds 
Reproducing sound to sound identical to reality is a somewhat fraught issue. It isn’t about distortion, frequency response or dynamic range - we can manage those to less than detectable levels. Where it goes wrong is that the sound field is messy, and we don’t capture anything like all of it. Stereophonic sound - with two channels is a very poor approximation to reality.
You hear the sound field around you with a remarkable amount of detail - the shape of your head and shape or pinnae affect the frequency response of the sound depending upon it direction, and when you move your head you sample the sound field at different angles and locations. All of this allows you to build a much richer idea of the sound field than a simple two channel recording can encode.
Next, a person speaking emits sound into a space, and the sound is emitted from a range of parts of their body - mouth mostly, but there are deep sounds directly from the chest as well. Further, again the shape of the person affects how the sound is radiated into the space - it isn’t equal in every direction, and thus the sound field in a space depends upon the person’s shape, and the space they are in.
So eventually, the reverberant field in a room is filled with a very messy result, and your ears sample that result. A recording with one or two microphones and a couple of speakers is simply unable to reproduce that same sound field, and you will be able to tell.
Worse, sound recordings of people use a lot of tricks to enhance the voice - many microphones have a proximity effect, where when you get close, their frequency response changes dramatically - enhancing the bass, and leading to the well known deep voiced DJ sound. Recording a voice to sound natural is a skill in itself. And natural depends upon the space the recording is made in - a recording samples not just the voice, but the reverberant field of the room it is recorded in. Another reason a reproduced voice can sound wrong is that you hear some of the characteristic sound of the recording venue in the room in which it is replayed - and it doesn’t belong, and you can tell.
This isn’t to say that these issues can’t be solved. But it isn’t trivial. One technique that does work is a binaural recording. But a recording that requires you to wear headphones to sound real isn’t what was wanted.