EP3819904A1 - METHOD OF DETERMINING THE SPEECH IN A WEB-RTC AUDIO OR VIDEO COMMUNICATION AND/OR COLLABORATION SESSION AND COMMUNICATION SYSTEM