To judge from your first sample of a piano piece the unwanted noise is not the 'self noise' I'd expect from any microphone, which would approximate to 'white noise', and is the wideband natural background electromagnetic radiation of our environment (think the rushing sound of a waterfall). The USB circuitry itself could be noisy. I've looked at your Audio-Technica spec and it's not clear whether the mic internally itself (rather than the USB interface) is 'balanced' or 'unbalanced', and the latter are generally pretty noisy and much less well screened from electromagnetic interference. Your noise sounds to me like a lowish frequency 'buzz' from an AC electricity main to me, or at any rate a harmonic of same, and could even be getting in from anything nearby connected to mains electricity, not just your recording equipment. Fluorescent lighting can be noisy inside the human hearing range, and some cheaper LED lamps can be really awful, and even interfered with my kitchen radio on FM, although better brands today are much less susceptible. Anything with a motor can also generate mains noise, such as a refrigerator, although this will generally cycle between 'on' and 'off'. A 'stud wall' (lightweight wood and plasterboard) between rooms won't efficiently block electromagnetic 'leakage' from a device close by in the next room, a brick / block construction provides much better better electromagnetic isolation. As others have said, you're faced with a process of elimination. I would even start by turning all the lights off! Tip: to really get a fairly accurate idea of a mic's self noise, bury it in lots of the heaviest, deepest upholstery you can possibly lay your hands on (I use a tight roll of offcuts from a high quality deep pile carpet, and then some, use the quietest possible space you can find in your home (if not your studio, maybe an airing cupboard??) trip lightly away on slippered feet and hit 'record' in Audacity. Don't move and stop breathing (only joking). You'll soon see the level of noise when you review your recording and up the magnification on the resulting waveform, and if you amplify it you can then easily distinguish between mains hum / buzz, the lived in environment around the mic, or the generalised 'rushing sound' of natural background radiation, which even a passive balanced mic will exhibit to some degree, along with all the 'active' electronics it feeds into. Finally, the rough 'n ready but effective way to 'bury' unwanted noise is to get as close as you can to the desired sound source WITHOUT actually overloading the mic beyond its designed maximum spl, i.e. the 'hottest' possible signal, rather than applying lots of gain to bring a weaker signal up to level, then record. Electromagnetically induced background noise is often a constant; it won't get louder unless you amplify it with unnecessary gain, so a 'hot' signal will basically 'swamp it', as well as assisting the biological property of 'noise masking' which is a poorly understood natural artefact of the way in which human hearing evolved (put crudely, the brain - within reason - stops paying attention to a soft noise when a much louder one is present, and this effect is very obvious in your sample as the piano gets louder, it's next to impossible to honestly say that you can still discern the unwanted hum as a discrete and irritating noise in its own right). It is all too easy for the audiophile to get distracted by the Holy Grail of technical perfections, rather than what one is really seeking: the artistic effect of the performance