There has been a longstanding bug in Chrome which caused the echo cancellation applied to the microphone input to ignore all sound produced by the Web Audio API. This was a major incompatibility and motivated some interesting workarounds. The issue got finally fixed.

However Chrome's new implementation differs again from any other browser. It will now use any audio produced by any tab as the reference sound for the echo cancellation algorithm. Firefox for example only uses the audio of the current tab.

The spec doesn't explicitly mention how the echo cancellation should be applied. Any browser might change it's algorithm at any point.

Let's learn how to use browser automation tools like Playwright to verify the browser behaviour to make sure any future changes to the algorithm will not surprise you or your users.