Captions, Transcripts & Audio Descriptions

Why Auto-Captions Must Be Reviewed

Auto-generated captions produced by Zoom, YouTube, and Box are typically 80–90% accurate under ideal conditions. That accuracy level is not sufficient for WCAG 2.1 AA compliance, which requires captions to be an accurate text alternative to the audio.

Common auto-caption errors include:

• Proper nouns — names of people, places, and institutions are frequently misrecognized
• Technical and discipline-specific terminology (e.g., "econometrics," "HECVAT," "FERPA")
• Numbers, dates, and statistics
• Cross-talk and overlapping speech
• Non-standard accents and speech patterns

Three Types of Accessible Video Content

Understanding what each type of alternative content is — and when it is required — helps you plan before recording.

1. Captions

   Synchronized text that appears on screen during video playback, representing all speech and relevant non-speech audio (e.g., "[applause]," "[music playing]"). Captions are required for all video content that includes audio. They benefit deaf and hard-of-hearing users, non-native speakers, and anyone watching in a noisy or quiet environment.

2. Transcripts

   A full text version of all audio content, typically presented as a separate document or text block below the media player. Transcripts are required for audio-only content (e.g., podcasts) and are strongly recommended for any video with significant spoken content. They also support search indexing and note-taking.

3. Audio Descriptions

   A narration track added to video that describes important visual-only information — actions, on-screen text, scene changes — that is not conveyed through the existing audio. Audio descriptions are required when meaningful information is conveyed visually without any spoken explanation. Example: a demonstration where an instructor silently clicks through a software interface while students are expected to follow visually.

Zoom Recordings

Zoom is the primary platform for lecture capture and meeting recordings at CMC. Follow these steps for accessible output:

1. Enable transcription before the meeting starts. In your Zoom web settings, go to Settings → Recording → enable "Audio transcript." This must be on before the meeting begins — it cannot be added retroactively to a recording.
2. After the meeting ends, edit the transcript in Zoom Studio. Log in at zoom.us → Recordings → select your recording → click the transcript. Review and correct errors, paying particular attention to names, technical terms, and numbers.
3. Download the corrected VTT caption file if sharing externally. If you share the recording outside of Zoom (e.g., posting to Canvas or a website), download the corrected VTT file and attach it as the caption track for the video.

Zoom AI Companion is available to all CMC enterprise users and generates automatic transcriptions for meetings. As with all auto-generated captions, review accuracy before publishing or distributing the recording.

YouTube Captions

YouTube automatically generates captions for uploaded videos. These captions must be reviewed and corrected before the video is made public or shared with a class.

1. Go to YouTube Studio and select your video from the content list.
2. Click Subtitles in the left menu.
3. Under the English language row, click the pencil icon to edit the auto-generated captions.
4. Review each caption line. Pay special attention to names, technical terms, and numbers.
5. Click Save when corrections are complete.

Do not rely on YouTube's "Auto-sync" feature as a substitute for manual review. Auto-sync improves timing but does not correct word errors.

Box Video Captions

Box is CMC's institutional file storage and sharing platform, and its video preview supports subtitle and closed-caption tracks. Box does not generate captions for you — you supply a caption file alongside the video.

1. Create or obtain a caption file. Box reads WebVTT (.vtt) caption files. If your video was recorded in Zoom, download the corrected VTT file from the Zoom recording (see above). Otherwise, generate captions and correct them before continuing.
2. Name the caption file to match the video. The caption file must share the video's exact filename, followed by the language code — for example lecture-01.mp4 pairs with lecture-01_en.vtt.
3. Upload the caption file to the same Box folder as the video. Box associates the two automatically based on the matching filename.
4. Verify in Box preview. Open the video in Box, click the CC icon in the player controls, and confirm your caption track appears and is selectable.
5. Review the captions on playback. Confirm timing and accuracy, paying particular attention to names, technical terms, and numbers, before sharing the Box link.

Live Event Captions

Live events such as all-hands meetings, town halls, and institutional announcements have their own captioning requirements:

• Zoom live captions: For any Zoom-hosted live event, enable Zoom's built-in live caption feature before the meeting. Go to the meeting controls → Live Transcript → Enable Auto-Transcription. Participants can choose to display captions on screen.
• CART for high-profile events: For commencement, board presentations, all-campus town halls, or other events where accuracy is critical, consider Communication Access Real-time Translation (CART) — a human captioner who produces near-verbatim captions in real time. CART accuracy far exceeds auto-generated captions. Contact ITS to discuss options and lead times.

Contact the Help Desk at help@cmc.edu to discuss captioning options for an upcoming event.

Captioning Checklist

• All videos shared publicly or posted in Canvas have captions.
• Auto-generated captions have been reviewed and corrected for accuracy.
• Audio-only content (podcasts, audio lectures) has a full written transcript.
• Any video where meaningful information is conveyed visually without spoken explanation has an audio description track or equivalent text alternative.
• Caption files are in a standard format (VTT or SRT) and are properly synchronized with the video.