First of all, the whole process is described in the iBooks Asset Guide, section "Read Aloud Books".
Second, note that SMIL/Media Overlay is supported by iBooks ONLY for Fixed Layout ebooks. No official support for reflowable ebooks. (See also: https://github.com/pettarin/rb_smil_emulator )
Basically, you need to do the following:
- Assign an id to each XHTML element (usually, a text fragment, either a
<p>
or a<span>
) - Create the SMIL file, associating each id from step 1 with a time interval of the associated audio file, containing the narration of the corresponding text (this is the time-consuming part)
- Embed the SMIL file into your EPUB3 container, and add it to the OPF manifest
The Media Overlay specification is at http://www.idpf.org/epub/30/spec/epub30-mediaoverlays.html
You might want to start by looking at some samples:
- FXL samples:
- http://www.pubcoder.com/en/download/ (download sample EPUB, look for the page "Readaloud")
- http://azardi.infogridpacific.com/resources.html (several FXL samples)
- Reflowable samples:
- https://readbeyond.it/ebooks.html
- http://azardi.infogridpacific.com/resources.html (look for "A Christmas Carol")
EDIT: full disclosure: I am the Head of R&D at ReadBeyond.