Same problem. Microphone working as the VU meters respond accordingly; BUT the whole capture face fails in the end.
It would be helpful if the failure response told the user WHY the capture failed ie. at what stage the failure occurred. Was it the audio part, too much background noise, not loud enough or the video part as in the background being to bright, the foreground being too bright, enable to recognize facial features like eyes etc.
It's up there with Microsoft's highly informative... "An Error Has Occurred" message.
Now i have to use a process of elimination to determine if it is the webcam quality, the lighting setup, the audio surroundings etc I'll do it later.