When you talk, part of what you are hearing is the vibrations through your facial bones. When you hear others (or a recording of yourself), this component is not present.
If you record your voice on a high quality recorder, what you hear from the recorder is what other people hear.
There is also direct neural inhibition of your own voice. The middle ear muscles contract and the olivocochlear bundle is active, and both serve to reduce and at least the former will also filter the sound.