Web Speech Recognition on MediaStream

Calling SpeechRecognition.start automatically requests microphone permission, but is there any way to start SpeechRecognition on a custom MediaStream (e.g. from getDisplayMedia)?