An expensive microphone and a decent quite room to record in. Most microphones are aimed at voice chat and other such tasks which do not require high quality audio. As such they suffer from bad frequency response and signal to noise, especially in the high frequency range. Although you can remove the noise to some extent you cannot recover high frequency information as that was never recorded. A decent microphone should also have a good cone of attenuation which you can aim at your face to cut out most background noise rather than sampling every sound in the room like an ear or video camera.