Live transcriptions for your meetings!
Users of the Zoom @ YorkU service can benefit from real-time, live automatic speech recognition transcripts. While it uses the same Otter.ai machine-generated transcription, this feature is separate from the automatic transcription service available for Cloud Recordings.
While live transcription is an amazing addition to the Zoom meetings toolset, if your meetings require a higher degree of accuracy, and/or are required to conform to a standard for Accessibility, then you may still need to work with an outside provider who can guarantee much higher accuracy for your meeting.
Use of any translation services for Zoom will also need to be procured through an outside provider.
High accuracy in live transcription is always preferred. Usually human-generated outputs are the most ideal, and yield accuracy into the upper ninety percentages. This accuracy requires a massive force of human input that is unsustainable due to scheduling, resourcing and costs to employ such a force. Machine-generated output, on the other hand, can be accessed at any time, for a much lower cost and can reach a much larger group, on-demand. While the accuracy is better now than it has ever been. It isn't quite at the same level as a human-generated output.
Due to the artificial intelligence (AI) that the system uses, the accuracy will be below what a real person such as a CART stenographer or digital note taker would be capable of. To ensure you get the most accurate output possible please consider the following variables:
|Speaker clarity||Ambient noise||Multiple speakers|
|Microphone quality||Speed of diction||Overlapping Speaking|
|Specialized, industry-specific language||Mixing of Languages||Language accents|
The video covers both the account configuration, and enabling the function in your meeting(s).
Is this the same as Closed Captioning?
Somewhat. By virtue of the name "live transcript", the system is listening to, and writing down what it hears into a text file. A conventional Closed Captioned file will have timestamps on what is being said so they will play back properly each time the video is played.
Can this identify each speaker and add their name to the captioning so we can see who said what?
No. The AI is listening to the audio stream, and turning what it hears into words. While this service is incapable of performing this task, there are other third-party services that are pay-for-use or subscription based, that will do this. One example of this is Intact Streamer (streamer.center).
Can you make the accuracy any better?
Unfortunately what you see is what you get. There is no toggle switch on the administrator's side that can tweak how the service hears your meeting; and how it then transcribes what it hears.
Accuracy will get better over time as the service matures and the developers work with the system to improve the quality and capabilities of the service. These efforts will be through the Otter.ai service that Zoom uses for their transcription offerings.
What languages will work with Live Transcription?
At the moment, English is the only language that will work with this feature. We will update the community if the system's capabilities extend into other languages.