So let's analyze the low frequency cutoff:
At the input we have a high-pass filter made by C2, R2 and R3. For the audio signal R2 and R3 are in parallel so they are combined 50 k Ohm.
The cutcoff frequency is at
\$f_c = \frac 1 {2\pi RC} \$
For C = 0.1 uF and R = 50 k ohm that gives \$f_c\$ = 32 Hz
That's a bit high, I suggest that you replace C2 with a 1 uF capacitor which will lower \$f_c\$ to 3.2 Hz, that's low enough.
At the output something similar can happen depending on what you connect to the output. If you would connect to an input with 50 kohm input resistance then again you would get a cutoff point at 32 Hz.
However, most line inputs have a much lower input resistance though, in the order of 10 kohm. That will move the cutoff point to a much higher frequency and you will definitely hear that, the 32 Hz at the input is still " OK-ish" but much higher than that and sound will suffer (as you have experienced).
Again I suggest increasing the capacitor's value. I would make C4 10 uF, yes that's 100 times larger than you have now!