Multiple headset intercom (conference call sim)
ManAtWork
Posts: 2,180
For my flight simulator I'd like to build something that simulates the cockpit intercom system. This is something that basically does what Zoom does when we have a live forum meeting. Multiple persons (in this case pilot, co-pilot and ATC over the radio) have headsets on. The sound of all microphones that are above a certain threshold are mixed together and transmitted to all earphones. The mics below the threshold are muted.
There are possibly ready-to-use solutions avauilable, but AFAIK either for studio application or aviation and therefore too expensive. At first, I planned to implement this with analogue amplifiers. But I think the P2 is the better choice. It's much easier to add bandpass filters and noise so it sounds like a real avionics radio. Yes, I want it to sound bad on purpose, hifi is too easy. 
I know there are Audio accessory boards but they have seperate jacks for microphone and headphones. I need 4-way connectors for headsets with microphone included.
As I don't need high quality I'd use the internal ADCs of the P2 with a simple OPA as pre-amplifier. And IIRC headphones can be driven diretly with DAC pins if multiple pins are connected in parallel, right?
Has anybody done something similar? I know, mixer panel software with multiple sound cards in a PC could also do this. But small PCs (NUC, RasPi) have only one audio jack...
Comments
Preamp may not even be needed, I remember the first P1 board with a electret micro directly connected to the P1 and you could hear the birds singing.
I have one problem.
Imagine you have 4 intercoms on I/O pins and you digitize the waveforms. Now if you mix those 4 channels (just + the waveforms together) you might end up with values 4 times the maximum of your DAC. how to deal with that? if you divide the values by 4 then you won't clip but the volume will be 1/4 ..... how is this done?
Just clipping/dividing could work in many cases. Otherwise there's compression/soft limiting, wherein if the output exceeds some threshold, the volume is turned down (and slowly back up as the signal stops exceeding the threshold)
Yes exactly, intercom and radio always use something called "dynamic compression" which is basically a control loop that keeps the amplitude nearly the same. This can sometimes lead to funny effects. The voice of the speaker stays the same but background noise from the engine and wind get softer or louder in the opposite direction when the speaker modulates his voice. This also means that emphasis/stress is completely futile.Therefore, there's a rule that if you want to put emphasis on a word over the radio you have to say it twice, twice!
Mix at 32 bits. Then you can have a global gain variable that gets decremented automatically when the mix signal value approaches the threshold (eg. 32000) and then autoincrements (very) slowly. Multiply the mix value by the gain variable.