Don't mess with the capacitor if you have distortion.
Here's the "circuit" from your first link:
"TIP" and "RING" are the speaker outputs. "RING2" is ground.
The two 1k resistors form a voltage divider with the 100 ohm resistor connected to ground ("RING2.") The values given reduce the speaker signals by a factor of 10. That is, a speaker signal of 100mV is only 10mV when it gets to the capacitor. You need this because the speaker output signal is much larger than the normal microphone signal.
Distortion means the signal going in to "MIC" is too high. The solution is to use larger resistors in place of the 1k resistors suggested in the circuit. That will lower the voltage of the signal going to "MIC."
If you said "distortion" but meant that low frequencies are missing from the recorded audio, then you used the wrong word.
If low frequencies are missing, just say so.
The cure for missing low frequencies would be to use a larger capacitor.
A 1 microfarad ceramic capacitor would be simplest.
If you only have electrolytic capacitors, then the plus terminal should go towards the microphone. The microphone line has a DC voltage on it that is used to power the microphone capsule.
For various reasons, I had a look at the spectrum of the recorded audio using the Google "loopback dongle."
The 100nF capacitor causes considerable loss:
Red is the earphone, blue is going to the microphone.
Some experimentation showed that it takes a 4.7µF capacitor to flatten the frequency response:
The 50Hz spike is from my test setup - too much power line hum getting into things.