Sc1 2 1 Video Only Alternative

Please note, this is a draft

Description

This test checks captions in video elements without a track element. There are three different situations when a track element is not required

• Video is a media alternative for text and labelled as such
• Audio describing video-only content
• Video contains open captions

This procedure describes those situations

Background

• Understanding 1.2.1 Audio-only and Video-only (Prerecorded)

Assumptions

• Test device renders HTML5 video (phantomJS doesn’t render)

Test properties

Test procedure

Selector

Test mode: automatic

video element not containing a track attribute

Step 1

Test mode: manual

• Check if the main content contains text (hl: this could be an extra, automatic, step)
• IF the main content doesn’t have text.
  • GO TO step 2
• IF the Video is labelled a media alternative for the text.
  • IF the Video contains more information than the text
    • RETURN SC122media-not-alternative-for-text
  • ELSE
    • RETURN SC122media-alternative-for-text

There is currently no technology to make the label programmatically determinable.

hl: is this - in theory - not possible using aria-labelledby or zria-describedby?

Step 2

Test mode: manual

• Listen to the audio
• IF the audio is describing video-only content
  • RETURN SC122-video-only-added-audio

QUESTION: how can we automatic determine video-only content if audio is added?

Step 3

Test mode: automatic

First check if the video doesn’t contain text at all.

• Check if any of the video frames (may use samples nth seconds) includes text
• IF none
  • RETURN SC122-video-no-open-captions

Step 4

Test mode: automatic

• FOR EACH spoken audio fragment
  • FOR EACH of the corresponding video frames
    • IF the corresponding video frame contains text
      • NEXT audio fragment
  • (NO text found in all frames for this audio fragment) = RETURN SC122video-all-dialogs

Probably more forgiving: 80% of spoken audio fragments should include text?

Step 5

Test mode: manual

• FOR EACH spoken audio fragment
  • IF the content of the spoken text has NOT the same meaning as the caption text
    • RETURN SC122-video-caption-meaning

Step 6

Test mode: automatic

From this step we will check if non spoken visual information captured, captions are more than subtitles.

Check to see is a speaker is presented in a fixed frame with unchanging background (‘talking head’ video).

• FOR EACH frame (between intro frame and end screen)
  • IF the frame contains a head similar to the previous frame.
    • RETURN SC122-video-captions-provided

NOTE: not sure if we should add this step, but if this step can be done fully automatic with frame scanning, we can skip additional manual steps.

Step 7

Test mode: manual

Check for each sound if it is important and has descriptive text

Step 8

Test mode: manual

Check for each background fragment if it is important and the corresponding frame has a captioned text

Step 9

Test mode: manual

Check for each speaker fragment if its frame has a captioned identification

Step 10

Test mode: manual

Check for each emotion if it is important and the corresponding frame has a descriptive text