[Athen] Zoom and captions

[Robert Beach]

Hello folks,

You all may remember that I posted asking about how to turn on auto speech recognition in Zoom. My colleague and I have played around with it. Well, here are the results of our experimentation.

1)         Unless I missed it, Zoom does not offer real-time auto speech recognition. I personally do not feel this is a bad thing. Too many people think they can turn on ASR and they have met the need for captions in their live presentations. This is not at all true. For a real-time presentation/meeting, you need a live captioner to correctly caption as you go.

2)         Zoom does offer auto captions for recordings after they have been completed. This is a good way to get a draft of the caption transcript. Here are some pointers for this process.

A)        The feature must be turned on in your profile. This must be done through the web profile since the feature is not in the desktop app. The feature was already turned on in my account but was not on in my colleague's account. I'm not sure what the default setting should be.

B)        I'm not sure if the option is available in the free accounts since I am using a professional account.

C)        You must record to the cloud, not to the local machine.

D.        Apparently, you must stop the recording and end the meeting before the system starts generating the transcript.

E)        Once the transcript is finished, Zoom will email you and you can log into the web account to download either the video and/or the transcript. The transcript is a .VTT file.

F)         We uploaded the video and transcript to Panopto and we are quite happy with the results. The syncing was very good. Of course, there was some editing that needed to be done with the transcript, but it was very good.

G)        The transcript from Zoom indicated who was speaking automatically which was very nice.

H)        I did try sharing my screen with computer sound to record what the screen reader was saying. Unfortunately, the transcript did not pick up any of that. The screen reader speech has to be manually entered. This is still better than manually typing the entire transcript. <smile>

3) We did try loading the Zoom video into Panopto's auto speech recognition system to compare. Here is what we found.

A)        The accuracy was not as good as the transcript from Zoom. It did not take long to determine we would prefer the Zoom over the Panopto transcript.

B)        Panopto's ASR did catch the screen reader's voice, but it did not recognize much of the speech at all. We would still have to manually enter most if not all of the screen reader's speech.

C)        Panopto did not automatically mark who was speaking like Zoom did. This makes sense and was not a surprise.

D)        Panopto did sync the speech with the video very well.

My take-aways:

1)         For real-time captions, use a real-live captioner.

2)         Use Zoom to record and generate the transcript for the captions since Zoom does a very good job.

3)         Upload the video and transcript into Panopto and edit as needed.


Robert Lee Beach, Assistive Technology Specialist
Kansas City Kansas Community College
rbeach at kckcc.edu<mailto:rbeach at kckcc.edu>
913-288-7671
I am currently working from home. Please leave a voice message and I'll return your call as soon as possible. Email is probably the quickest way to reach me.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman12.u.washington.edu/pipermail/athen-list/attachments/20200714/eb83f9ca/attachment.html>