I don't follow what distinction you're making here. Stallman said Please do not distribute my speech in any other format. That means he's not ok with you distributing his speech using, say, H264.
No, he's not. But it sounds like he would tolerate H.264 plus Theora, if the Theora version was always provided to people that the H.264 version was provided to.
> it sounds like he would tolerate H.264 plus Theora, if the Theora version was always provided to people that the H.264 version was provided to.
I don't agree that this is a reasonable interpretation of the text. I really don't think he could have been much more clear: he is not ok with this.
For audio codecs, Stallman could not be more explicit. He's not OK with using non-Free codecs, and makes no exception for where multiple codecs are offered:
> I insist on using the formats of the free software community: Ogg Vorbis or Ogg Speex format for audio, and Matroska VP8 (Webm) or Ogg Theora for video. Please do not distribute my speech in any other format.
Clear and categorical. He is similarly explicit on the topic of video codecs:
> Please do not ever broadcast or publish my speeches in formats that are not good for free software.
There's no wiggle-room here, he uses the word 'ever'. From a little later in the document:
> you must use only Ogg format or Matroska VP8 (Webm).
Again, no wiggle-room. There is no implication of an exception permitting the use of non-Free codecs when multiple encodings are offered.
With all of that said, the FSF say here [0] that they're ok with H264 as a fallback. That page was published just 4 days ago, so perhaps they've reconsidered their position. See also the (currently empty) HN thread on that page. [1]
edit For what it's worth, in Feb 2016 Stallman was still going with the hard-line, see his 2016 talk in Switzerland at [2]