Producing Captions

Captions enable people who are deaf or hard of hearing to follow media content, although they can be beneficial to everyone. Captions are a text display of dialog and narration and are written in the same language as the audio (subtitles, on the other hand, are a translation of the audio into another language). They also contain important non-speech information, such as sound effects or speaker cues.

Captions can be open or closed:

• Closed captions can be hidden and revealed by users, typically by operating a button or menu on the player's control bar.
• Open captions are continually displayed and cannot be turned off.

There are three ways to present captions to viewers:

• Pop-on captions appear in discrete blocks and usually contain one to three rows of text. They are normally created for pre-produced material, and used for the majority of captioned online videos.

  Example:
  

• Roll-up captions scroll up onto the screen, one row at a time. These are normally created for live programming, but can also be used for pre-produced material.

  Example:
  

• Paint-on captions are text that appears to quickly unfurl onto the screen, one character at a time, as the data are received. Typically, when one row of captions finishes painting onto the screen, the row scrolls up and a new row begins paint onto the screen.

  Example:
  

Production workflow for pre-produced captions

A typical caption file contains timing information and positioning codes in addition the text representing the audio track. Timing information indicates when each caption should appear or disappear from the screen; positioning codes indicate where on the screen the captions should appear. There are a variety of do-it-yourself tools available for creating captions for pre-produced video and audio clips.

Basic workflow for creating pre-produced captions:

Below is an image of a caption editor showing how captions have been formatted and timestamped. Each caption is assigned a start/display time. In this editor, captions that are not assigned end/erase times will simply be replaced when the next caption displays.

Caption quality

Always provide the highest-quality captions that convey 100% accuracy. Keep the following points in mind when writing captions:

• Ensure that there are no spelling errors. This includes the names of characters or speakers.
• Use conventional grammar rules. After end punctuation (period, question mark, exclamation point, etc.), always begin a new caption block.
• Do not edit unless you have a specific reason to do so (e.g., to achieve a specific reading level). Fillers such as "um," "ah," etc., can be deleted to save reading time unless doing so alters the representation of the speaker.
• Do not censor: captions should reflect the words that are spoken in the audio track.  If objectionable words are used in the audio, the captions should show those words.  If the audio is edited to obscure a specific word or phrase (e.g., "bleeped" audio), the captions should reflect the fact that a word or phrase has been obscured.

See resources for writing quality captions for more information.

Styling captions

Most caption-creation tools let authors style captions in a number of ways: e.g., adding color to the text or background, specifying different font faces and sizes. However, support in browsers and other media players for styling information is inconsistent and is at times unreliable. If the media is being produced for a specific player, style the captions to that player’s capabilities. Otherwise, rely instead on a player’s default presentation style (usually white characters on a black box).

Below is an image showing bold text added to captions in a caption-authoring tool.

And here is the WebVTT markup for that caption…

1
00:00:00.670 --> 00:00:06.680
The genome is a storybook that's been edited
<b>for a couple billion years.</b>

…and the TTML markup for the same caption:

<p xml:id="s_1" begin="00:00:00.67" end="00:00:06.67">
The genome is a storybook that's been edited<br />
<span tts:fontWeight="bold">for a couple billion years.</span></p>

Automatically generated captions

Automatically generated captions should never be used as the sole method to produce captions, but they can be a part of the production workflow. See the discussion about automatic captions for more information.

Resources for writing quality captions

The importance of presenting users with high-quality, accurate captions cannot be overemphasized. Use the guidelines below to help create captions that are informative and easy to read.

• DCMP Captioning Key
• DCMP Captioning Tip Sheet
• Captioning FAQ