Cleanout Starts on September 20
As we all know, Stream is in the middle of a migration from its old platform based on Azure storage to ODSP (OneDrive for Business and SharePoint Online). All new Teams meeting recordings are now stored in ODSP and Microsoft is preparing for the next phase, which is to migrate videos stored in Stream classic to ODSP. A bunch of work is going on to prepare the way, like a new web player for video content (MC261352, last updated August 10, due for a complete deployment by the end of September) and a new way of viewing the transcript of Teams meeting recordings (MC274185, July 30). In other words, things seem to be progressing nicely on the Stream 2.0 front.
Clearing Out Old Video Transcripts
That is until you read MC279467 (published August 20. At first glance, the text seems inconsistent with the generally positive progress of Stream. The notification says that starting September 20, Stream users will be unable to access (view, search, or edit) the transcripts automatically generated by Stream for:
- New videos uploaded to Stream. However, transcripts will be generated once the newly uploaded video receives one view.
- Older videos which have not been uploaded or edited (for instance, to trim a video) in the last 3 months (U.S. East datacenter region) or 6 months (for other regions hosting Stream services like the other U.S. regions and EMEA).
In other words, Microsoft is stopping the automatic generation of transcripts for new videos and removing transcripts previously generated for some older videos. This doesn’t affect the storage of videos; it is all to do with their transcripts, which are a significant accessibility feature for people to follow what’s happening in a video. Figure 1 shows an example of the transcript viewed alongside a video recording for my “Talking Teams” interview with Rish Tandon, Microsoft VP for Teams Development.
If someone views an older video, Stream will regenerate the transcript. The same will happen if an owner edits video details and chooses to regenerate the captions (Figure 2) which make up the transcript.
Microsoft will not remove transcripts for videos when:
- The transcript has been edited (for example, to correct some of the phrasing generated by automatic transcription). Although the automatic transcript is OK for most videos, it can have problems with idioms, unclear speech, and when people talk over each other.
- Transcripts are uploaded manually.
- Videos are active. To qualify, videos are uploaded, edited, or viewed in the last six months.
Interestingly, closed captions remain available for all videos.
Why is this Happening?
It’s reasonable to ask why Microsoft is cleaning out old Stream transcripts. Cynics will say that it’s a cost-saving measure to drive profits and point to the recent announcement about increased monthly fees for Office 365 plans. I think the answer is a little more prosaic. Microsoft is preparing for a migration. The nature of migrations is that they are often painful, complex operations. Removing data that doesn’t need to be migrated makes sense, so the possible reason is that Microsoft wants to clean out transcripts for videos which haven’t been watched or edited in a while so that they don’t need to migrate the data to ODSP. Given that transcripts probably need some massaging on ODSP to be indexed and become searchable there, this is a plausible driver.
Microsoft isn’t saying why the clean-out is happening but given that the migration is expected sometime early in 2022, it really doesn’t matter. After all, if someone notices that a transcript is missing for an antique video, it’s easy to regenerate it.