Electronic – Noise canceling while recording from mobile on PC

My guess is that it may be the cheap Crossplay supply, which is coupling (capacitive from the mains side) it's switching frequency onto the signal/PC/mains ground.
I would try using another (preferably better quality) supply of the same voltage rating (with same or better current rating) to see if things improve. Or you could try a ferrite bead clamped on the cable of the current supply if you have one handy.
If you have any means of isolating the Crossplay from the PC (e.g. little 1:1 audio transformer or similar between Crossplay/PC) then that should also do the trick if it is as suspected.

EDIT - guess 2 :-)

If it's not the Crossplay supply, then it's almost certainly the PC supply. In this case I think the easiest way to sort it is probably to buy the little isolator you link to, rather than trying to filter. These things are very handy to have lying around if you are into audio in any way.
In the meantime you could maybe try things like plugging the PC supply into the same socket as the Crossplay if it isn't already, or using shorter/thicker cables.
Audio noise problems can be quite difficult to diagnose, but IME (I have a small recording studio) an isolator will usually fix things, and often the easiest option.
One more thing - if your Crossplay happens to have a differential input (unlikely I guess) then use that.

Assumption: The desired end-product is some form of hardware identification / copy-protection dongle for software to run on a smartphone.

Going from simple to more reliable, here are two options:

1. Sound ping-back: Pure analog, no microcontroller

   • Dongle consists of a circuit as described below, connected to 3.5" (or 2.5", whatever) TRRS headset connector, to be plugged into smartphone's headset socket.
   • App on smartphone outputs an audio tone in a series of bursts (varying on-off duration) on one predefined audio channel (Left or Right) of headset.
   • Dongle circuit harvests power separately from each headphone channel: Consider a Schottky diode bridge rectifier fed by the audio signal, and with a zener diode as a shunt regulator... one on each headphone channel.
   • If audio signal is present on both channels (i.e. normal music playing, for instance) the net voltage between the harvested channels will be low to none - so the rest of the device simply will not power on, or will do so intermittently (variation of audio between channels)
   • If on the other hand, a steady sound signal exists on just one channel, i.e. a voltage difference is established between the two harvested channels, this will be used to power a simple audio-frequency oscillator of any sort, emitting a predefined frequency (the infamous 555 timer won't work, too power-hungry... perhaps a low power variant), which outputs to the microphone channel of the headset connector.
   • The smartphone app receives microphone-input, and detects whether the incoming signal is the frequency expected and none other, within analog component and design tolerances. Also, it can check whether the mic input signal exists only as long as the audio out signal is present - note the variable tone bursts mentioned above.
   • If yes, then dongle is valid, app is good to go.
   • Note that this form of dongle is trivial to crack, by splicing in an oscilloscope or even an audio recording device, between dongle and socket.

2. Unique ID Challenge-Acknowledge via audio. Uses a low-power microcontroller.

   • Dongle again harvests power from headphone channels to power itself, as above
   • App on smartphone repeatedly outputs a unique numeric sequence as a DTMF tone with a start marker, on one predefined audio channel (Left or Right) of headset.
   • Microcontroller's ADC is used to sample incoming DTMF, to extract the numeric sequence once the start marker is identified.
   • If sequence is not a known valid one as programmed into the microcontroller's code, nothing is done.
   • For a valid sequence, a corresponding numeric sequence is generated by some predefined transform, as complex as desired.
   • This resultant sequence is output as DTMF tones by the microcontroller, back through the microphone channel of the headset jack.
   • Smartphone application receives this DTMF sequence and validates against the transformation algorithm. If the result checks out, the dongle is valid, and the app is good to go.
   • A low-power microcontroller such as the Texas Instruments MSP430 family is recommended for this kind of design.
   • Open source DTMF generation and detection source code can be found for many microcontrollers, and of course for Java or whatever other language the smartphone app is to be coded in.