Using YouTube Captions & Transcript to Find Content in a Long Video

on February 23, 2018

This week, I was pretty pumped to see that PASS published Erin Stellato (twitter) and Dejan Krakovic’s (linked in) excellent session on Query Store to the public.

I attended this session live at the conference, and I remembered that there was one specific part on memory limits in Query Store that I wanted to watch again. I’d jotted down a note about it, but things were flying by and my note was gibberish.

The recording is 75 minutes long, and it’s a great session, but I wanted to just find this one part at the moment. However, I couldn’t remember if the comment was early or late in the session.

Transcripts to the rescue!

Transcripts are kinda hard to see in YouTube, and I only discovered them when I was playing around with YouTube in the process of captioning my own videos.

To open transcripts, I had to click on the ‘…’ in my browser, then select open Transcript. That gets it open in a pane of its own:

 

If you’re on mobile, you may well get a different UI experience and have to tap elsewhere. My general practice is just to rub my forehead against the phone until it works.

Click in the transcript box, and use Find in your browser

The transcript for YouTube videos is generally automatically generated, so you may need a little creativity or luck here.

I knew I wanted to find a reference to ‘memory’, and it seemed like a good word to search on because it doesn’t have a lot of homonyms. I guess it could be auto-transcribed as ‘mammary’, which would be hilarious, but in this case the computers got it right.

The whole transcript loaded in the page for me, so I found all 18 memory occurrences, and was able to page through them.

Use the timestamps to get to that point in the video!

The timestamps helpfully showed that the portion of the video I wanted to watch is around 37 minutes in, and I was able to go straight there.

So, what is this part of the video about?

There is a question from the audience about the performance impact of running Query Store.

Dejan gives an answer that points out that the overhead is often minimal, but you may run into special situations if you have a workload that generates a lot of unique query plans (such as a non-parameterized workload). I won’t spoil it, watch it for more info – for the portion on the safety memory limit, watch until  38:42.

Thanks to Erin Stellato, Dejan Krakovic, and to PASS for publishing this video!

I loved the session live, and it’s great to have it online to reference and share with others.

You can watch even more free videos from the PASS Summit 2018 right here.