My question:I have a feeling the circuit I came up with might not be the best solution for the problem I'm trying to solve. I hope somebody can review the circuit. And maybe recommend a better solution or point me to some useful information.
The goal of the circuit:I would like to have an indication about the average noise level inside a room.
No specific frequencies. Lots of sound or a higher sound level needs to result in more voltage.
Requirements:(1) All signal processing inside the analog domain, because of privacy concerns.
(people don't like it when there is a device inside the room with a microphone inside, connected to a MCU and attached to a transmitter). When the audio signal processing is done inside the analog domain it is easy to prove no sound can be recorded and transmitted.
(2) Low power, because the dive is battery operated. (not that crucial because the noise sampling circuit will be turned on an interval)
How I imagine the circuit to work:The MCU turns on the circuit;
Delay x seconds; (Sampling time)
The MCU measures the circuit output voltage using the ADC;
The MCU turns off the circuit;
The circuitThe signal from the mic passes trough a DC-blocking cap.
Is amplified and rectified using a half wave rectifier (I don't need the full wave, so this simplifies the circuit).
Is buffered and passed to an LOW-pass filter to create an increasing voltage.

This circuit wont work, because the op-amp is single supply and U2A (pin3) is referenced to ground.
Possible Solution:
Adding a DC-offset to the amplifier works, but now I need to remove this offset, using an differential amplifier or digitally by taking a adc-sample just after the circuit is turned on and substract it from the last adc-sample.