Microsoft Stream is Microsoft’s default, web-based video player for video files stored in OneDrive or SharePoint, including Microsoft Teams recordings.
Among other features, it offers an AI-based speech-to-text captioning service for transcribing audio.
Automated Audio Transcription
It’s crucially important to review the output transcript for accuracy, as all AI transcription services make mistakes. Microsoft Stream has an interactive transcript editor that lets you easily modify the transcript to correct those mistakes. Once corrected, the entire transcript file can be downloaded for additional offline analysis.
Stream’s transcription service is only made available on files it thinks are video. Audio-only formats (like .wav
or .mp3
) will first need to be coerced into something Stream considers to be video. See Coerce Audio to Video below.
Transcribing Video
Stream will not generate captions for files larger than 1 GB. If your file is larger than 1 GB, please use video editing tools to split it into multiple files.
Upload your video file to OneDrive or a Sharepoint Site.
If clicking the file name in OneDrive or Sharepoint does not open the file in Stream:
hover over the file name
click the 3 dots on the right
hover over Open
click Open in Stream
Click
Video Settings
on the rightExpand
Transcripts and captions
Click
Generate
Select the primary language spoken in your video, and click the foreground
Generate
buttonDepending on the length of your video, automatic transcription may take a long time (hours) to complete. It’s safe to close the browser tab at this point; transcription will continue on Microsoft’s servers in the background.
Coerce Audio to Video
If you have an audio-only file to transcribe, you must manipulate the file into a format that Stream considers to be video before Stream will present the transcript generate button. The easiest way to do this is to use the free VLC video player app to copy your audio file to a new file in a specific container format that Stream considers to be video. As of January 2025, Stream will consider any file using the Matroska (MKV) container format to be video, even if it doesn’t contain a video track.
VLC works on both Mac and Windows, and is available in both the CIT Self-Help app for college-managed Macs and Company Portal on college-managed Windows devices.
To coerce an audio file into an MKV container file:
This process requires you to navigate your computer’s file storage and locate files. If that feels overwhelming, we recommend watching these Windows or macOS file browsing training videos on LinkedIn Learning before continuing.
Launch VLC
Click
File
➡️Convert / Stream…
Click
Open media…
, then locate and open the audio file you want to convert. Your selected file should now be visible above theOpen media…
button.If you previously created a profile for this purpose in VLC, select it from the dropdown of profiles, and skip to step (5). Otherwise:
Click
Customize…
On the
Encapsulation
tab, selectMKV
On the
Audio codec
tab, selectKeep original audio track
(you can ignore theVideo codec
andSubtitles
tabs)Click
Save as new Profile…
, give this profile a memorable name like “Microsoft Stream Transcribe Audio”, and clickSave
Click
Apply
Click
Save as File
, clickBrowse…
, select both the output folder and desired output file name (the extension should default to.mkv
), and clickSave
Review your
Convert & Stream
window, which should look something like the following. When you clickSave
, VLC will process your input file using the selected profile, and save the output to your chosen destination file.Upload your output file to OneDrive or Sharepoint, then follow the Transcribing Video instructions above to open it in Stream and begin transcript generation.