Electronic – Line Level to Mic Pad Working Randomly With Phones

I'm trying to build my own headphones to mic pad. I want to connect the input of the pad to a phone's headphones conector and the output of the pad to a second phone's mic connector. Both phones feature a TRRS 3.5mm audio jack (4 contacts, CTIA standard). Here's the circuit:

                     C1
    +Line level in --||----R1----+-- +Mic level output
                       +         |
                                 |
                      +----R2----+
                      |
    Ground (input)----+--------------- Ground (output)

I found the schematics in this page

I've measured the headphones output signal in a couple of phones. Each brand seems to have different amplitudes, almost nobody follows line level standard. I measured amplitudes ranging from 500mV to 1.8V peak to peak.

I've chosen R1=20K, R2 = 1K, so the output amplitude of my pad should be 1/21 the input (23mV to 85mV).

This pad works fine when plugged to any device that only has a mic input. But with phones it's a bit more complicated. Phones already have a built-in mic, and there is some mechanism in place to detect whether or not an external mic has been plugged to the audio jack (e.g. a handsfree). So with phones and tablets, this very same pad works only sometimes: when it works, the pad is detected as an external mic, and it displays the "headset+mic" icon in the status bar; when not working, the phone switches to the built-in mic, and the "headset with no mic" icon is displayed in the status bar (as if only a TRS connector were plugged in). It looks like a random issue, and I don't know why, because I used exactly the same phones and tablets for my tests. My pad is well soldered btw.

And now here comes the odd thing. If I add a 1V DC offset to the pad output then it works always. No random malfunction any more.

Could it be that the unbiased pad output signal (which is centered around 0V) is detected as "ground" by the phone and it switches to external mic? Or is it a bad choice of resistor values?

UPDATE: it could be the case that my voltage bias circuit is providing aditional impedance and that makes it work. I'll play with a potentiometer in R2 and see what happens.