Can i do a full-text search where each word has location metadata?


I'm trying to build a text search from voice to text, each word should have a metadata position something like start: 54s; end: 57s.

Metadata shouldn't be indexed but only returned with results. Is that possible? I'd appreciate some advice on how to do it.

(Mark Walkom) #2

You'd need to figure this out before sending to Elasticsearch, there's nothing native that could calculate this.


Thank you for the replay, currently I can get text and position from audio.

My current idea is to make a regular text search then use a custom highlighter to get the text position. With the position query an SQL database for a location in audio.

I'm wondering if I could do the whole thing natively with Elastic

(Mark Walkom) #4

Elasticsearch cannot do the sort of merge/join with the words and the timings, you may need to do that in your client.

But you could put both sets of data into Elasticsearch then you've got a quick data store to leverage at least :slight_smile:

(system) #5

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.