Content Markup or Presentation Markup for Audio Rendering of MathML

Greetings everyone!

My aim is to create a parser that generates a text version of formulas (MathML) in technical documents, to feed a TTS engine.

What would you use for audio rendering of mathml?
Content Markup or Presentation Markup?

Seems to me that Content Markup has much more to offer for audio rendering, however, most of the math applications or conversion tools only create documents with presentation markup.
Most people use simple editor's that create presentation markup (usually in XHTML+MATHML format to publish in web).

The problem is that a parser to have some utility has to think of what people use, and pratice tells me to go for the Presentation Markup, but Content Markup is much more suitable for audio rendering.
And transforming Presentation Markup into ContentMarkup doesn't seem a good idea to me, since presentation cares about visual rendering and content cares about the structure and the meaning of the formula, and this is what matters to a blind user, for example.

W3C MathML Spec and MathML Handbook from Pavi Sandhu, aren't very helpfull in this area, i only see one unique reference saying that Content Markup is suitable for Audio Rendering. But seems to me that is shooting in the dark since most people use Presentation Markup. Or am I wrong?

Is Content Markup only to be used as a standard markup language between applications? Or can it be used to publish technical documents too, with the possibility to be usefull to render it in a TTS engine?

Thank you for your attention,

    Helder Ferreira
    Laboratory for Speech Synthesis, Electroacustics, Signals and Instrumentation
    Faculty of Engineering University of Porto, Portugal