Only institution-level Administrators have access to this ASR option.
It is important to note that the feature described on this page applies only to automated transcripts generated through EchoVideo provided transcription services; it does NOT apply to auto-ingested Zoom transcripts.
EchoVideo has partnered with several transcription service companies to provide an automatic speech recognition (ASR) service. The ASR service sends media to the transcription provider to automatically create machine-generated transcriptions for video/audio media.
EchoVideo provides an ASR Settings configuration, that allows you to:
- Configure ASR Language Settings to identify the primary language spoken in the media being transcribed, or to set auto-detection of the primary language.
- Auto Send ASR to Captions setting for automatically applying transcripts to the closed captions track for media being transcribed (this article).
The Auto Send ASR to Captions option allows you to tell EchoVideo to automatically push the media transcript to the closed captioning track, if it meets the confidence score threshold you set. Typically closed captions are more accurate than media transcriptions, however where the transcripts are determined to be very accurate, there is no reason not to use them as closed captions.
EchoVideo has had the ability to manually push transcripts to closed captions, but this feature allows administrators to have that done automatically.
Because some transcripts are NOT as accurate as others, and because accuracy matters for accessibility reasons, the auto-push feature provides a "confidence score" threshold that administrators set. This threshold must be met or exceeded before the auto-push happens.
If the media already has a captioning track, the auto-push feature will NOT override the existing captions. Replacing those will have to be done manually, likely through the EchoVideo Transcript Editor. If the media receives closed captions from a captioning provider AFTER it already has captions generated from a transcript, the newer captions from the captioning provider WILL be applied to the media.
The section at the bottom of this page provides more details on what a confidence score is and what it means.
NOTE: Enabling this feature or changing the confidence score does NOT back-apply the feature to media with existing transcripts. Only those transcripts received AFTER enabling/changing these settings will be subject to your configuration. In addition, as mentioned immediately above, if the media already has a closed captions track, regardless of how it got there, it will NOT be replaced by transcripts via this feature. Those media must have CC tracks replaced manually if desired.
To configure automated closed captions from transcripts
- Log into EchoVideo as an administrator.
- Click the Settings icon in the upper-right corner of the screen.
- From the Settings menu, select Institution Settings.
General settings appears and Basic Info is selected by default.
- Click Features.
- Scroll down to the Transcripts and Captions section of the page, shown in the below figure.
- Click the ASR Settings button, identified in the above figure.
The dialog box that appears contains both the Auto Send ASR to Captions setting discussed here and the ASR Language Settings options.
- Enable the Auto Push ASR Transcript to the Closed Caption Track toggle.
Enabling the toggle activates the confidence score field as shown in the below figure.
- Change (or accept the default) Required Confidence Level Percentage for automatic push to closed captions. The default is 95%. See the section below for more information about confidence scores.
- When finished, click SAVE.
All transcripts received from this point forward will automatically be pushed to the closed captioning track IF the confidence score given by the transcript provider for the whole transcript meets or exceeds the percentage given in these settings. This change WILL NOT be back-applied to any media that already has automated transcripts. Furthermore, if the media ALREADY HAS closed captions (regardless of how they got there) those WILL NOT be replaced by auto-pushed transcripts.
See Automatic Transcriptioning Service or ASR for additional details surrounding the transcription service. See Enable/Disable Closed Captioning for more information on disabling (at least temporarily) your Closed Captioning service if appropriate.
What Does the Confidence Score Mean
In machine-generated transcriptions, each word is given a "confidence score" based on how certain the machine program determines it was in selecting the correct one. Lower confidence often indicates speech that was garbled, difficult to understand, lower in volume, or other reason it may not have been interpreted properly by the automated program. The scores are presented as a percentage and as such range from 0 to 100.
EACH WORD in the transcript has a score. The Confidence score used for the auto-push to closed captions setting is the TOTAL AVERAGE SCORE across the entire transcript. So that if you set the confidence score for the feature to 90%, the total average score for ALL WORDS in the transcript must meet or exceed 90%. Meaning that there may be transcribed words or sections of the media that are lower than that, but that the majority of the words in the returned transcript have been determined to be accurate.
Keep in mind that the score given to the transcript elements is provided by the same program that did the transcribing. So there is no certainty that the assessment is completely valid, although it is often close. And you may find that if a word is used repeatedly but has been transcribed wrong (using a close substitute), the program may mark that term with a high confidence score because it provides it consistently. This is common if the transcript is in American English but the speaker is British; the transcription program may not get the word right, but it gets it the same every time.
What you may want to do is to spend time reviewing auto-applied closed captions from transcripts, to determine if those normally returned are accurate enough for this use. At that point, you may want to set the confidence score for the auto-push feature high enough that only those transcripts that are very good get applied as closed captions. Others can be manually edited, and once verified, pushed to the closed captioning track. In fact, if the transcripts you receive are generally "good enough", you can still have the transcripts applied as closed captions, have the transcripts edited for further accuracy, and then re-pushed (manually) to the closed captioning track.