There are some very useful and important use cases for allowing speech recognition in a tab that has lost focus; including accessibility, dictating notes while viewing other tabs, and command-and-control.

To alert users when speech recognition is active, Section 4 already states "User agents must only start speech input sessions with explicit, informed user consent" and "User agents must give the user an obvious indication when audio is being recorded."

Thus I propose we delete the sentence "To minimize the chance of users unwittingly allowing web pages to record speech without their knowledge, implementations must abort an active speech input session if the web page lost input focus to another window or to another tab within the same user agent." This sentence only adds use case restrictions without adding any real protection, particularly when considering that a malicious web page can enable recognition without providing any visual indication on the web page.  The real protection is through the "obvious indication" provided by the user agent.

Also, specifications such as WebRTC do not place similar restrictions on web page focus.