You're probably not gaining anything by using digital. In general the tricky thing is clocking the audio in-and-out at a constant rate while tying to run the rest of your program. (i.e. Soundcards have their own clock and input/output buffers so the CPU can multitask.)
and then send it out to an amplifier via PWM (where I have seen several examples but am not understanding, what's going on).
Since the Due has a DAC, you can use "normal" PCM. (If you don't know how that works, the [u]Audacity website[/u] has a good introduction to how audio is sampled & digitized.
...merge them (which is my first problem)
...1. How to I "summarize" two analog audio-signals
Audio mixing is done by addition (summation). Analog mixers are built-around [u]summing amplifiers[/u]. You mix digitally by summing the samples. i.e. Sample 1 from microphone A + sample 1 from microphone B = output-sample 1, etc. (44,100 samples per second in the case of CD audio.)
But, you do usually have to scale the data... If you are adding a 12-bit sample to another 12-bit sample you get a number that's too big to send to your 12-bit DAC. So in reality, its more of an averaging process (i.e. divide the samples in half before summing). In the real world, it's often a weighted average because all of the audio inputs have their own volume control and they are not necessarily mixed/summed equally. (Most audio editing/processing software uses floating-point so you don't have to worry about bit-depth except at the input/output.)
- Is there a rule of thumb what resistors/capacities are required between Arduino and Mic as well as Arduino and Speaker/Amplifier.
Line level inputs are typically 10K - 100K. The line inputs on a soundcard or the input to your powered computer speakers are usually in that ballpark.
A series DC blocking capacitor makes a [u]high-pass filter[/u] (DC is zero Hz so it's blocked by a high-pass filter). Given a 10K typical minimum impedance, 1uF is a typical capacitor value to block DC while allowing the lowest audio frequencies through.
Note that if the microphone shield is biased you do NOT want a capacitor because you do not want to block the bias. If the shield's output is not biased, you need to add a
[u]bias circuit[/u] because the Arduino can't read the negative-half of the AC audio signal.
You should add a series capacitor keep the Arduino's bias out of the amplifier. (And, you do need an amplifier because the Arduino can't directly drive a speaker.)
(oro even three in case there comes the day, when I want to do a recording)?
To record digitally, you simply save the samples in an array or in a file. (Usually in a file, since even at a low sample rate of 8kHz, there are 8000 samples per second.)
which is a beep-overlay. Probably not the perfect low pass filtering constellation yet.