I had the same thought but the voice is already in the original recording. It's the same phenomenon like the voice saying "Jenny" you discovered in my previous clip. It has a characteristic that is different. Generally the voices in the carbon clips are distorted and spiky. You have to make them intelligible by applying reverb and filters. The other voices are very clear, like gained with a microphoneand in fact I remember those voices ONLY from my experiments with direct voices where I used a condenser electret microphone with a self designed preamplifier.

But here the crucial question comes up again; if we record direct microphone voices, do they really com THROUGH the microphone or are they using another way? It's strange since I had planned to do investigations on direct microphone voices before the carbon thing came up again.

If we are suggesting that natural sounding voices can manifest without a microphone the question arises how they do that? I am hypothesizing that this is the point where Hodowanec comes into play again. What if they are coming in through a capacitor?

The capacitor C2 is a polarized cap like Hodowanec used them and it is in the feedback network of the OpAmp. Thus electrical fluctuations in the capacitor will scale up at the output of the OpAmp. With P1 set to a low value you almost have the Hodowanec circuit here.

Yes I see, a good idea. What I have done today was to connect the + input of the OpAmp directly to the middle of the power supply with no signal input for the OpAmp. Then I looked for pk modulation. In a 2nd run I enlarged the capacitor upt to 2200uF. In both cases I had no pk modulation. To be sure I need to make longer test runs of course.

Have a good night, Jeff!