AudioConverterRef with different number of channels

I have an audio format with 4 channels that I want to convert into 2 channels format. kAudioConverterChannelMap can only discard the extra inputs:

When In > Out, the first Out inputs are routed to the first Out outputs, and the remaining inputs are discarded.

Is it possible to specify channel map with kAudioConverterChannelMap to merge 1 + 2 channel into one and 3 + 4 channels into the second of the output format?

If not, what should I use for this conversion?

Solution 1:

AFAIK Apple does not provide an API that reduces the channel count of an audio stream, apart from discarding extra channels.

Why? I guess because how you do this is such a personal choice that if they did think about it at all they may have come to the conclusion that there was no one approach would make everyone happy and so best instead to choose the approach that would make everyone angry: discard.

So what's the big deal? I think it's mainly to do with the "meaning" of the individual channels. Consider the "obviously simple" case of converting stereo to mono - it's surprisingly nuanced, depending on how the stereo audio was recorded. And that's in an ideal case where your 2-channel audio is actually declared to be "stereo" (e.g. via an AudioChannelLayoutTag like kAudioChannelLayoutTag_Stereo).

So even "tagged" multi-channel audio is terribly ambiguous, so the only person who knows how to interpret and reduce the channel count of your 4 channel audio is you.

That said, would it have killed them to provide a function that summed even channels with even and odd with odd using some Accelerate/vDSP array functions, like vDSP_vadd? Because that's probably what you want.