Many years ago at a NAMM show, I was listening to a mic (cardioid) on headphones at an exhibitor's booth. It sounded fine when pointed around the room, passed the key-jangle test, but when I listened to my own voice, it sounded thin. I looked at the preamp it was connected to and found the polarity (phase) switch was set to the reverse polarity position. I flipped it back where it should be and the mic sounded fine. I pointed it out to the guy manning the booth, who didn't seem to know what that switch did. I wonder how many other people listened to the mic that way.
The same thing occurs when monitoring through a DAW, though for a different reason. The delay (latency) going from the mic to the A/D converter, the guts of the software, the D/A converter, and back to your ears via headphones causes the headphone sound to be somewhat out of phase with the sound reaching your ear drum through the natural organic path. If they're close to being the same volume (at your ear), the 1/2 to 2 millisecond typical monitoring latency added to the natural sound creates a few comb filter notches in the important part of the vocal range.
When I've pointed this out in the past, I was surprised at the number of people who say "I've never noticed that." I believe that the reason why is that they have the headphone level turned up to ear bleeding levels so they can hear their vocal over the drums or guitar. The headphone signal dominates the sum of the two sources and the comb filter notches are small enough not to be noticeable. I tell them to start talking with the headphone level all the way down, then bring it up slowly while talking.