As someone new to Lucene, with a specific problem to solve, it is difficult to identify the appropriate Lucene feature to use. Reading various online posts, I see I'm not alone. I have a use case that I think this JIRA issue addresses; perhaps it will help refine what the issue resolution would do.

I'm indexing a lecture video transcript. I want to store the text of the transcript and timecodes of when each word occurs. I want to search the text of the transcript and return the timecode so I can display the lecture video from that spot.

Hi Peter,

I work on the same thing. You can get my code from here: http://speech.fit.vutbr.cz/en/software/speech-search (Lucene extension for bin sequences), there are also some testing data. Actually it indexes word confusion networks with scores of hypotheses, but of course it will work also for 1-best string transcripts.

That code runs behind this website: http://www.superlectures.com/odyssey/

It is few months old, so if you are interested, I can send you our current version.

Best regards,
Michal Fapso

If we want to implement something like LUCENE-4272, we would need this option. I don't think it would be too bad especially
now that term vectors use the same codec apis as the postings lists, e.g. hasPayload and getPayload is already there,
its just that today it always returns false.

See http://mail-archives.apache.org/mod_mbox/lucene-dev/201208.mbox/%3CCAOdYfZV_r6Ov8WFpAOyPid2r6MqR-_zx7kDmVvjucN1wcxD8Yw%40mail.gmail.com%3E

Closed after release.