You use the TrueDepth camera (dot projector and all the same used for FaceID) to map your pinna and then it will compute that somehow to tailor the response, coupled with how it already tailors the sound with Adaptive EQ (uses the microphones to hear how the music is playing in you ear).
They also mentioned AirPods are the most popular model lineup of headphones in the world. As a teacher of high schoolers, this is easy to believe.