Indexing an oral history interview: best practices?

Hello all; I am an oral historian, new to Omeka as a platform. My question is this:

After an oral history is complete, I have an audio file, a transcription doc file, and a doc containing an index pertinent to my projects and organization.

How do you recommend incorporating that index into the item? Options besides cutting and pasting parts of it into my "tags" field?

Thank you!

To clarify, the existing index may be in either a Word or Excel file. Words are associated with timestamps. I'd like the index itself (WITH timestamps) to be searchable by users and admin.

Am I following right that the index on the site ultimately needs to have three parts to it: 1) a reference of some sort to the oral history (which is now the separate index document for each item), 2) the word, and 3) the timestamp in the oral history where the word occurs?

So it might be like this as CSV/Excel?
history 1, "Jefferson", 2:13
history 2, "Jefferson", 5:21

If that's accurate, I think this could be done, though in a bit of a roundabout way. First, I'll simplify it to remove the requirement of the index including the timestamps.

In that case, you could put the relevant words from the index in as different Subject elements for each Item. The SearchByMetadata plugin to make those subject entries act as an index onto the Items.

Like I said, so far that ignores the timestamping. To layer that in, you could also create a new Element under Item Type Metadata called something like "Timestamped Index" (or whatever you want), and include the full Word-timestamp pair there.

So an item might have:

Dublin Core Subject: "Jefferson" (linked via SearchByMetadata to other items)
Oral History Item Type Metadata: "Jefferson: 4:20" that displays the timestamp of the occurrance

The upshot is that you'd have the word-level indexing as the Dublin Core Subject, and the word+timestamp level as something that's recorded as a separate metadata element. There's a little duplication there, obviously, but it seems like that'd get closest to having keyword indexing and the additional reference to the timestamp.