If you have a transcript, Kaltura makes it easy to synchronize it to the video in (typically) as little as 30 minutes with their machine alignment feature.

Caution

Captions created with this method *must be edited* to ensure that they accurately convey auditory information and appear at the appropriate times within the video.
To learn how to edit captions, *be sure to follow the link in the final step.*

DIRECTIONS

Part 1: Prepare the Transcript

In order to auto-synchronize or machine align a transcript, you first need a transcript. The transcript can be professionally made, manually created, or constructed from edited autocaptions and other machine generated text. However it is created, it needs to have the following qualities. Click on each one to learn more.

Accurate Translation of Audio to Text (Including Good Grammar)

Accurate Translation of Audio to Text

In order for captioned videos to be accessible, the captions need to accurately translate audio to text. This means *correcting any machine generated text* AND *ensuring that proper grammar is used*.

For example, let's imagine that a video clip containing footage of luggage on conveyor belts has the following spoken sentence:

"Most of the time travelers worry about their luggage."

The person speaking it voices the word "time" differently and pauses slightly after saying the word. This means that the following sentence is a more accurate translation.

"Most of the time, travelers worry about their luggage."

Without the appropriate grammar, the Deaf and Hard of Hearing audience members could be led to believe that the clip is about the number of time travelers who worry about their luggage instead of the amount of time that travelers worry about their luggage.

Simple edits like the one in the example may be easy to make after the transcript has been synchronized, but we still recommend that they be made before synchronization so the alignment tool may make better decisions on where to begin and end caption frames, thus improving the readability of captions.

Grammar Tips

Indicate interruptions with double hyphen (--) or emdashs (—) even when the speaker is interrupting their own thought.
Omit Ums and other verbal ticks unless they are purposely included to create a dramatic effect or personality.
URLs such as “www dot D-C-M-P dot org” should be written as it would normally appear: “www.DCMP.org”
When describing important sound effects, put them in brackets: [applause].

Removal of All Extra Text (Including Timestamps)

To get the best captions from your transcript, your transcript must not contain any text beyond the dialog or narration, speaker identification(s), and sound effect description(s). This means that document titles and footers and other details that may be included in a professional transcript will need to be removed. This also means that the timestamps in autocaptions will also need to be removed.

If you need to remove timestamps, you can use the desktop version of Microsoft Word and the process below to achieve this task rather quickly.

Open the transcript in Microsoft Word on the desktop (i.e. not in your browser).
Open the "Home" tab/ribbon then open "Replace." (Shortcut key: Ctrl+H or Shift+⌘+H)
Scroll to the last time stamp in the document and copy it.
Paste the timestamp in the "Find What" field.
Replace each number with ^# (example: 5:05 will be ^#:^#^#)
1. For .srt files, you can paste the following into the "Find What" field: ^#^#:^#^#:^#^#,^#^#^# --> ^#^#:^#^#:^#^#,^#^#^#
2. For .vtt files, you can paste the following into the "Find What" field: ^#^#:^#^#:^#^#.^#^#^# --> ^#^#:^#^#:^#^#.^#^#^#
Click "Replace All" and allow all timestamps in the document to be replaced.
Repeat steps 3 through 6 for any remaining timestamps.
(If you are working in a .srt file) put ^#^#^#^#^p in the "Find What" field, click "Replace All" and repeat this process, decreasing one ^# each time (so ^#^#^#^p then ^#^#^p, then ^#^p).
Put ^p in the "Find What" field.
Add a single space with the spacebar in the "Replace With" field.
Click "Replace All."
Do the following as needed.
1. If the words at the start and end of paragraphs become one word, press Ctrl or ⌘+ Z to undo and try steps 8 through 10 again.
2. If the text is not in a single paragraph, click “Replace All” a second time.

Screenshot showing the process for using the Replace Tool.

New paragraphs for each speaker

Putting each speaker's part into a new paragraph may not be necessary if all speakers end on a complete thought that is marked with appropriate punctuation. However, we still recommend this practice in order to help Kaltura ensure that multiple speakers do not appear within the same caption frame.

Please note that, if you identify speakers at this stage, you should not use colons (:) in the identifications as Kaltura may automatically remove them during its alignment process.
(We don't know why this happens.)

[Most Important] Saved as Plain Text or TXT File Format

Save as Plain Text or TXT File Format

Kaltura requires that all transcripts submitted for synchronization or alignment be saved as a Plain Text or .txt file.

Open the transcript in a document editor.
Open "File" then select "Save as" or "Save as a Copy."
Set the file to be saved to an easy to find location (such as the desktop).
Set the file type or file format to "Plain Text" or ".txt".
Click "Save" or "Okay."
If asked about conversion or encoding, set encoding to "Other encoding" then "Unicode (UTF-8)" and "Allow character substitution."
1. This step should prevent Kaltura from replacing unsupported punctuation with boxes or other strange characters.

Screenshot of File Conversion dialog in MS Word with the encoding and character substitution parts highlighted.

Part 2: Auto-Synchronize (or Machine Align) the Transcript

Caution 2

Since this was written, we have found that using Kaltura to synchronize a proofread transcript back to the video can negatively effect the caption quality. Kaltura uses AI to apply autocorrections as it synchronizes. Because of this, it may be more beneficial to use YouTube to synchronize transcripts and then bring the caption files it generates back into Kaltura. The following is as much of a guide as we have for synchronizing transcripts with YouTube at this time. A new iteration of accessibility guides is being explored and further updates will come after that.

Auto-Synchronize Transcript in Kaltura (ppt)

Click to Skip to Additional Tips

Complete transcript preparations.
Upload your video to Kaltura and wait for it to finish processing.
- See Clemson Online's Kaltura page for uploading directions.
In "My Kaltura" or "My Media," find and click on the desired video.
Open "Actions" and select "Caption & Enrich."
In the region that opens below, set the drop menus as follows:
- Unit: Clemson [4299903] REACH-ASR
  - (This menu is only in "My Media" / MediaSpace--i.e. not in Canvas)
- Service: Machine
- Source Media Language: English
  - If other languages become available, then choose language spoken in video or audio.
- Feature: Alignment
Click "Upload txt."
In the "Upload Attachments" dialog box, click "Select File" then select your .txt file.
Fill out the remaining fields (optional) and then click "Save."
Click “Submit” then wait.
(Optional) Scroll up and expand the request details to see expected turnaround time.
Complete the caption editing process.

ADDITIONAL TIPS

If the directions above are not accurate because Kaltura has changed, see Kaltura's REACH - Ordering Captions For Media Per Entry Guide.
When proofreading the transcript, use double hyphen (--) or emdashs (—) to indicate interruptions—even when the speaker is interrupting their own thought.
Before submitting the transcript for synchronization or alignment, remove colons from speaker identifications or Kaltura may remove the entire speaker identification during the alignment process (We don't know why this happens).
Putting each speaker's part into a separate paragraph may help Kaltura improve readability by preventing words from multiple speakers from appearing in the same caption frame.
See the DCMP Captioning Key: Speaker Identification and DCMP Captioning Key: Sound Effects and Music for formatting recommendations.
- Important: At this time, it is recommended that speaker identifications avoid colons. Kaltura's synchronization tool currently removes speaker identifications in the following formats-- Name: and NAME:. We don't know why.
If you plan to autocaption or use speech recognition technology to machine-generate a transcript, you can do the following during the recording to improve accuracy:
- Use a lapel microphone, headset microphone, or other external microphone that can be placed near your mouth.
- Use a wired microphone when possible as they are more reliable for capturing audio. USB microphones tend to have better quality audio than 3.5mm audio jack microphones.
- Put a windshield (thin foamy material) over your microphone.
- Choose a recording location that is quiet and a safe distance from white noise generators like fans and air handlers.
- Speak clearly: Enunciate, but don't over enunciate.
- Speak slowly but not too slowly.
- Speak with enough volume to be heard clearly but don't speak so loudly that you overpower your microphone.