http://www.multiupload.com/FJ82JS1ID5
With the new added resistor (red dots only connected) you can see in the video loud and clearly how the ring gets detected which swings from a thread til 10cm with this improvised coil - with both cables at the FETs gate. The circuit now is much more sensitive and louder. Perhaps the basis of T2 needed some additional current, but I don't know why this has to be te one from the speaker.
Directional works only with grounded reflector. I made some fancy tests with my mini-tesla-coil plasma ball which got detected from 1m.
And I tested an op-amp LM358 connected at the output of the FET - louder but important analog signals were missing so I still use the transistors.
Perhaps antenna- and out in the wild-tests tomorrow or in the next few days.