Some stills from a test video working with separation of elements of speech and gesture. The funnel stops me speaking and removes facial gestures from the image. The gestures are in response to a block of text that derives from the ‘voices’ sequence.