The main issue you'll face is the line-level outputs have 2.2 uF capacitors.
As a quick test, I just ran this on a Teensy 3.2 and audio shield.
Code:
#include <Audio.h>
AudioSynthWaveformSine sine;
AudioOutputI2S audioOutput;
AudioConnection c1(sine, 0, audioOutput, 0);
AudioConnection c2(sine, 0, audioOutput, 1);
AudioControlSGTL5000 audioShield;
void setup(void) {
AudioMemory(12);
audioShield.enable();
audioShield.volume(0.8);
Serial.println("setup done");
sine.amplitude(0.9);
sine.frequency(0.5);
}
void loop(void) {
}
I measured the headphone output and the line-level output. Both show a nice 1/2 Hz sine wave!
Notice the first waveform is centered around 1.5V. That's the headphone output. The way it works is the signal is DC coupled, but the headphones get a "virtual ground" output, which is actually 1.5 volts DC. So from the headphone's point of view there's no DC voltage. But if you try to use this as input to some other system, you have to deal with the fact that the headphones don't have normal ground.
The line-level output has a 2.2 uF capacitor in series. This forms a high-pass filter, together with the resistance of whatever load it's driving. In this case, the only load is a 10X oscilloscope probe, with is approx 10M ohm. The high-pass filter frequency is 1/(R*C*6.28). So for R=10e6 and C=2.2e-6, the high-pass filter frequency is 0.00723 Hz. That allows the 0.5 Hz signal through quite nicely.
However, if you were to feed the line-level signal to something like a modular synth input, which is normally 100K ohm impedance, then the highpass filter becomes 0.723 Hz. It will greatly attenuate a 1/2 Hz signal, and it'll badly affect 1 Hz, and even have some small impact on 2 Hz.
You could modify the audio shield, to increase the capacitor, or just replace it with a wire. If you go for DC coupled, you'll have to deal with the DC voltage. If you go with a huge capacitor, you should consider the length of time for the DC voltage to stabilize to zero volts. Even 2.2 uF takes many seconds with the 10M oscilloscope impedance.