I would like to use the Web Speech API in Google Chrome to recognize speech coming from a HTML5 <video> on a web page. Is there a way to use the Web Speech API with an audio input coming from somewhere else than the user's microphone?
On MDN (https://developer.mozilla.org/en-US/docs/Web/API/Web_Speech_API) it says:
Speech recognition is accessed via the SpeechRecognition interface, which provides the ability to recognize voice context from an audio input (normally via the device's default speech recognition service)
But it does not clearly specify if this audio input can come from somewhere else than the user's microphone, or if it must come exclusively from the user's microphone.
I'm going to answer to myself. Maybe it can help others who had the same interrogation.
After more research I saw that an issue was filled in the Chronium project in 2014 to make this possible: Issue 408940. It was even added to Chrome Platform Status as a proposed feature.
Unfortunately, the implementation was never finished and the functionality was removed. (Removed in March 2017, see the Issue 701229 in the Chronium project).