Thanks to Syd, Laurent, John, and others. As I noted in my previous post, I don't think that the <graphic> element may be an approriate practice in this regard and I only point to it as a mechanism which is designed in TEI and I thought the lack of a similar mechanism for audio type files. This seems also the case for using <facsimile> element which (as I noted) is specially for graphic objects containing the source text. The <binaryObject>, although its definition may seems compatible with an audio file, and its content model may let us do that, but it seems specifically designed for a graphic type object (especially when we take a look at its special attributes). But, if we have no other option in hand it may be reasonable and somewhat preferrable to use it for inline audio which may be the case for some books in linguistics, phonology or so in the near future (I think that it may be better than adding a new element for myself).
I've also some problem with the <ptr/> practice. As I noticed from the guidelines it can be used if there exists some type of explicit referencing which is not the case here (am I right?).
Please let me know about the best practice you suggest in this regard. Also I'd be gratefull if you point to my mistakes if you think that I've mistaken (about your suggested approach or so).
Again, I want to recall the second case: the texts of great importance which may be a subject of teaching for some famous scholars which may be treated as a teaching of parts of the main text or some vocal comments or description or so about parts of text. Is it feasible to mark that part of text with <seg> element (if it's not marked with another element) and using corresp or ana attributes? Isn't that an abuse of these attributes? Is there exist any other practice? Does you think that the TEI may expand itself in this regard?
Again, I should excuse all of the list members because of my bad english composition and some silly questions.
Best regards, Saeed