Question on Speech to Text feature

[Collins DeLoach]

Is the "speech to text" and susequent "search within transcript" feature now available or is still in the works? I don't see it enabled on my existing videos.  Is there something else we need to do in order to make that work?

[#2]

You need to specify the language of the video for us to kick off automatic transcript creation. Right now we only support english and spanish. So if you are speaking a different lanague in the video it won't work. It normally takes about 1-2x the length of the video before the transcript shows up.

The transcript and deep search capability will only show up for users with E5, O365 Stream add-on, or Stream Plan 2. If you don't have one of those licenses you won't get access to use deep search or face timeline.

[#3]

Can you clarify that? No transcript at all without E5, add-on or Plan 2 or just not deep search /face timeline? I'm not seeing the transcript either despite having the option, matching the language requirements and waiting 2 hours after uploading a 7 minute test video.

[#4]

Did you on the video itself set the "Video language" to either english or spanish?

Without E5/Add-on/P2, we will generate an closed caption file for uploaded videos (if language is set to eng/sp). So after we generate it, and you refresh the player page, you could click the CC button on the player to see the closed captions. However you will NOT be able to see the transcript on the side of the video, search it or search within the transcript across Stream.

Only E5/Add-on/P2 will get the transcript on the side and deep search within/across the portal. (If you are a 90-day trial user we also let you get faces/transcript search).

Does the CC button show up for your video now? 

When you are editing a video in the options pane is there a "Download file" link in the captions section?

If your video is marked as eng/sp in the video language field and it's still not generating a closed caption file, then it sounds like we might have a bug. Let us know and we can see if we can troubleshoot from our end of it would be better for you to log a support case.

[#5]

I don't have a CC button and in the option pane there is no "download file" link. It is marked as eng/sp. Edit screenshot attached. I can open a ticket if you would prefer. Maybe it just hasn't fully rolled out to all tenants? We're on E1.

[#6]

@Brian Payst - I'm going to send you  a private message to get some more data from you about the video and your tenant so we can pull logs.

[#7]

I see the faces option (I turned it on) and I see the faces options under the video now.  However, the "search within trascript" option doesn't appear to be working.  What else needs to happen to have that option enabled?

[#8]

---

@Collins DeLoach wrote:

I see the faces option (I turned it on) and I see the faces options under the video now.  However, the "search within trascript" option doesn't appear to be working.  What else needs to happen to have that option enabled?

---

This is what's required... 

1. You set the "Video Language" to either English or Spanish on the video. We only support those languages right now.

2. You have to wait a while (1-2x the lenght of the video or so).

3. You need to be a user with E5/A5/O365 Stream add-on/Stream P2/Trial license. If not we won't show the faces or the transcript.

I'm guessing you have all of the above done already. If that's the case it's possible we weren't able to generate the transcript for the video.

Based off the investigation we did yesterday for the previous report of this, it looks like we can only auto-generate a caption for MP4 and WMV files. If it's something else we likely can't generate it at this point in time. I'm working with our dev team to understand a bit more so we can document specficially what's supported and what's not supported at this time. (Which codecs ,etc).

We are working to move to a new version of the transcript service / indexer that we use which will add more language support and support for more file types. That project is estimated to complete by end of this year.

Sorry for the incovience. We'll work on getting our documentation updated with more specifics on what file types work with transcript and which dont.