Crosstalk cancellation, or “Why Dr. Edgar Choueiri is my hero,” PART 1
In case you haven’t heard, there was huge news in the world of binaural audio this spring. Edgar Choueiri, head of Princeton University’s “3D Audio and Applied Acoustics” lab, announced a breakthrough software algorithm that solves crosstalk cancellation for binaural audio playback through speakers.
“Great!” you’re probably thinking. “But what the heck is crosstalk cancellation?”
Well, to wrap your head around crosstalk, first we need to check in on the basics of binaural audio, which I’ll go over in this post.
- Humans are able to hear and localize sounds coming from all directions because our brains can detect differences in those sounds between our two ears. The most crucial differences are time delay , and volume.
- For example: Imagine that a friend calls to you from across the street, in front of you and to your left. The sound pressure waves from their voice travel towards you, reaching your left ear first and your right ear a split second later. Also, since sound losses energy as it travels through the air, and your friend’s voice has to travel farther to reach your right ear, it is slightly weaker when it arrives. Your brain subconciously recognizes these differences, allowing you to locate your friend, even with your eyes closed.
- Binaural audio is recorded with a dummy head microphone (like “Fritz,” our Neumann KU-100), that has two omni-directional microphones embedded in its ear canals. This allows the mics to record and preserve the time/volume differences of sounds that reach its ears.
- Now here’s the kicker: To be able to properly listen to a binaural recording, you need to make sure that the signal from the left microphone goes only to your left ear, and that the signal from the right mic goes only to your right ear. Doing so allows your brain to decode the localization cues from the recording as if you were hearing them live, giving you the full “3D experience.” Of course, the easiest way to do this is with headphones.
- So here’s the problem. When you listen to any stereo recording through speakers, the left/right signal separation isn’t preserved. The signals from both speakers blend and interact with each other as they travel through the room, muddying the localization cues and altering the stereo image. This, my friends, is crosstalk. Sure, things that are panned to one side still seem to come more from the one speaker than the other, but it’s hard to pinpoint a specific location.
It’s not a big deal when you’re listening to a close-mic’ed, traditional stereo recording, that’s specifically mixed to sound good in this way. However, crosstalk can seriously mess up what makes binaural audio so beautiful. (I’m of the strong opinion that binaural recordings made with the KU100 still sound great over speakers, but crosstalk undeniably wipes out the 3D effect.) For that very reason, intelligent audiophiles have been attemping to find ways to minimize and cancel out crosstalk for years, but never very successfully. That is to say, it’s been done, but not without noticeably affecting the tonal balance of the recording, which sucks a lot of the fun out of listening to it.
But wait! There’s hope! Speaker-listeners and stereo-lovers of the world can rejoice, because thanks to Dr. Edgar Choueiri, binaural 3D audio is coming to a listening room near you.
More on that in Part 2 of “Why Dr. Edgar Choueiri is my hero.”
Before you read on though, check out this demonstration of an environmental binaural recording (turn up the volume until it’s about as loud as you can recall an average downpour to be). Listen with speakers and sure, it sounds like rain. But listen with headphones, and I bet you’ll duck when the umbrella opens!