Modern encoders (especially Opus) are indeed impressive at preserving the stereo image at high quality settings. If you are on a stable connection getting the full 192kbps+ stream, the phase error is likely negligible.

The issue is that we can't control the delivery.

Streaming platforms use adaptive bitrate. If a user's bandwidth dips on mobile, the player might switch to a lower tier where the model starts aggressively quantizing the Side channel (L-R) to save space.

Since the binaural effect relies entirely on that Side channel difference, I wanted to remove the variable entirely.

By generating it client-side with the Web Audio API, we get mathematical certainty regardless of the user's connection speed.