It's all about the audio chain, gain staging, the room and the talent (your voice). I haven't used the Yeti but my first guess is that by going from mic thru USB that the real problem is at the sound card level on your computer.
I would prefer to go into a nice pre before making the conversion from analog to digital, but assuming that Blue's engineers knew what they were doing, then my guess would be that your equipment either can't handle the signal well (sound card) or your levels going in are not set correctly (gain staging). This all assumes that the signal is good going into the mic (talent). And that the signal is the right type of signal going into the mic at the proper distance(mic placement and room). Experiment - if it sounds good it is good.

Phil

First, this is another side address microphone. And a lot of people just point these in the wrong direction. It's a simple mistake that sounds much like what you are providing. The acoustics of the room that you are in, it is actually what you are complaining about. There is nothing wrong with the microphone whatsoever. Utilizing compression... that's good. But with the compression also comes accentuation of the bad room sound. Much of this can be dealt with electronically, in the software, by re-drawing a compressor into a downward expander without compression. This is a tricky thing to do. Setting the proper threshold level is extremely critical. And it's this threshold setting that allows words to be heard and not necessarily the acoustics of the room. Sometimes this is referred to as noise gating but noise gating on spoken word actually sounds rather unnatural. Downward expansion can be set to a limit of how much downward expansion you want to reduce the deleterious acoustic effect of the room without slamming the door on the room. Or so to speak. These are engineering techniques that take some time and experimentation to get right. Sometimes you might find presets in your software indicating noise gating? Well, that usually sounds pretty awful on spoken word. Good on drums, snare, tom-toms, bass drum but not good on vocals. And I haven't seen any software yet that has good downward expansion with a decent GUI. And that's why it's tricky to create from a compressor algorithm. Nevertheless, it's something I always do especially for spoken word productions. It makes all the difference in the world. It's a startling difference for sure. And I would be happy to post some examples just to demonstrate what I'm talking about. It certainly isn't the be-all end-all with really awful sounding room acoustics such as yours. It would certainly though, be quite a help.

This technique of downward expansion also prevents the sound of gasping breaths, when the threshold is set just right. I have recorded thousands of commercials for multimillion dollar advertising agency, always utilizing this technique. I frequently utilize the downward expander after the compression. A lot of people prefer to use it before the compression. And sometimes I've also done that though, that's not my favorite way to do that. In fact I find it more difficult to do it that way. And that's because the compression actually evens out the gasping breaths making the downward expansion threshold easier to set. And this threshold control is not something you will find anywhere in the software. You have to have a compressor with a graphical user interface that looks like a grid. With no compression selected, the line in the grid will be at a perfect 45° right-hand angle. You create a point at around -10 to -20. Then you go to the lower left-hand corner and create another point. You then drag that point from the bottom left-hand corner toward the center of the display window. Everything below the upper point will now be going through some dynamically downward expansion. Expansion generally requires a fairly fast attack time and a release time around 75-150 ms. And you don't need much more than 10-15 DB of downward expansion for this to be extremely effective. Generally I downward expand by about the same amount of decibels as the amount of compression gain reduction I'm using. This provides for a very tightly controlled dynamic range for the spoken word along with a much more intimate and tighter sound. It's really amazing what else you can combine with the spoken word, once you've done this. Music and other background sounds will mix quite nicely with the spoken word track. There is a lot more that can go into this but that's the important basics.

My audio is always up when it's down right.
Mx. Remy Ann David

After listening on a better system I agree with Remy. It sounds like a poor room and phase issues caused by reflections to me.

Phil

i wouldn't hesitate to 'try' large diaphragm dynamic.

I got a track from a client who tried to record his vocals in his bathroom since that is where he usually sings in the morning when taking a shower. The sound on your example sounds like the same problems he was getting in his all tiled bathroom. He thought that by using the bathroom he would get some natural reverb. Best of luck!

The brain is the most powerful Digital Signal Processor. It lets you hear only what you want to hear. Unfortunately, when you really hear something the way it is, you become completely disillusioned and generally disappointed. I bet you never thought about becoming a Hollywood movie producer without a picture? As in a film made for the blind. Only sound. Then it becomes a completely different thing altogether.

One of the problems with the younger generations is that they never had the opportunity, in this world, to have not had television. In the mid-1970s, I was lucky enough to follow one of the last science fiction radio dramas. EG Marshall's CBS's Radio Mystery Hour. This was old-time, old-school, from before the television era. It was movies with a blown projector bulb. But it didn't matter. You could still see it. It was right there in front of you and in your head. I don't see sound any differently than those actual images. We can all see the acoustic aberrations without ever hearing them. And then when we do hear them, we're certainly not surprised. So we're not just recording engineers and producers, we are psychoacoustic construction workers.

Where's my hard hat?
Mx. Remy Ann David

I'm not a sound technician by any stretch of the imagination, but I thought that I would mention what I was able to do to get the "tin can" effect to go away for me:

1) The cable that came with the Yeti was the main culprit. Grabbed a (much) shorter USB cord from my camera and that fixed most of the problem.
2) Then I followed all the steps in the video below and it fixed everything else:

Just, sharing because a month of searching for solutions kept bringing me back to this thread. :)