In the UK, some of the debates in the Houses of Parliament are televised. On 18th December 2018, Jeremy Corbyn was filmed muttering something—which was interpreted by a Twitter user as "stupid woman"—to himself in response to a speech to Theresa May, the beleaguered British Prime Minister.
Later, in response to this Twitter accusation, Corbyn said that he had said stupid people, not stupid woman. A lot of media outlets said that several lip-readers that had viewed the footage thought that he had said stupid woman.
However, the other lipreaders that advised the Speaker of the House, who had to rule on whether Jeremy Corbyn had transgressed, said that it was not possible to tell which thing he had said.
If he did or did not say stupid people, how do we know? What is the LINGUISTIC evidence?
If we cannot tell, why is it that it is LINGUISTICALLY ambiguous and that we cannot tell?
An answer to this question would ideally involve some description of articulatory phonetics. (See, for example, StoneyB's comments below)
An explanation of visemes and how this concept bears on the question would also be very welcome.
Here is a video of parts of that original exchange: