Captioning HIDVL: Collection Assessment, Style Guide, & Workflow
Project Overview & Context:
For the New York University Digital Library with the Hemispheric Institute, I created an organizational style guide on how to appropriately caption Spanish or English language videos accurately. Only automated captions are available for over 1000 video performances in the Hemispheric Institute Digital Video Library (HIDVL). Accessible captions will raise awareness and recognition of these artists, allowing communities across the Americas to see themselves authentically represented, increase collection research use, and create an immersive experience everyone can enjoy.
Year
2024
Client
New York University & Hemispheric Institute
Project Phases:
1
Research & Assessment
Assess the collection & determine the main issue with the video itself and with the current captions.
2
Edit-A-Thon
I planned, executed, and led the Captioning HIDVL Edit-a-thon, during which the first draft of the style guide was tested to gain feedback on what could be improved.
3
Final Draft & Review
Using feedback from Edit-a-thon participants, make any necessary changes to the Final Draft of the Style Guide.
Research & Assessment
Systemic Elements
When it came to captioning videos, I took into account various systemic elements, including the tags on the previews of the videos on KONCH and the video metadata on the NYU Digital Video Library page.
For example, if we were looking at a video in Spanish, the metadata would help us find the artist's specific country of origin so we can accommodate the captions to that country's slang. If the artist were from Argentina, we would then look at Argentinian slang to make the captions as accurate as possible.
Using the tags KONCH had on the video previews also helps determine whether the captions were auto-generated or an AI transcript and whether they had already been precision reviewed.
Also, the title for the videos on KONCH was the last string in the URL that would lead you to the video on the NYU Digital Libary page.
Software
The captioning software used for these videos was KONCH. The software itself had some neat features that were useful when captioning, such as offering to summarize the video transcript to get an idea of what the video would be about.
But like with any software, KONCH possessed obstacles of its own, such as the following:
Timestamp creation issues
Adding more speakers than there was
Lack of slower speed availabilities
The image to the left showcases the KONCH captioning interface. Any words marked in red were usually words or phrases that the system was unsure if they were captured correctly.
Personal Findings & Video Captioning Strategies
My primary captioning approach was to time myself and see how long it would take me to caption a video compared to its actual length. I also recorded the speed at which I would listen to the video, as that affected the time it took me to caption. I then logged the main issue the video had, albeit if it was in both English and Spanish, and the speakers would go back and forth code-switching, or there was poor audio quality.
I also brainstormed different methods to implement to ensure an immersive and accurate experience and thoroughly researched different and practical approaches to captioning. Through this, I developed captioning strategies, which are as follows:
Using metadata before captioning can help give you a glimpse of what you will be captioning, basically giving you a heads-up of what to look out for.
Timing yourself can help you see how long it takes you, on average, to caption a video and determine how long a particular video may take you.
Using a slower playback speed can ensure that you’re captioning correctly, as it can be easier to hear the phonetic sounds of the words people are saying.
These tips and findings were also heavily shared and incorporated in phase two.
2. Edit-A-Thon
The project's second phase involved planning and leading an Edit-A-Thon event. During the event, I gave participants the first draft of the style guide and assigned them a video to caption. The event lasted an entire week.
I put together a Google folder that would house all the necessary information the participants would need, including:
Style Guide (Draft 1)
KONCH Manual & How to set up their account
Captioning Tips and Recs Infographic
Editing Flowchart
Common Issue Guide Map
Feedback Form
Shared Excel sheet with their assigned video, any notes they had, and their data on how long it took them to caption and if they finished captioning the video.
Part 2 of the Edit-A-Thon was the primary event day. We held a Zoom meeting, and I guided everyone step by step through what they were supposed to do. It was a four-hour meeting that included training on how to use the captioning software KONCH and if anyone who had any questions or needed any help could ask. I then held office hours for the rest of the week, offering guidance to anyone who needed it or had any questions.
Once they finished their video, they would fill out the feedback survey, which I would then use to edit and finalize the style guide.
3. Final Draft & Review
The project's final phase was more feedback-focused, as I used feedback from the edit-a-thon to determine any necessary changes to the style guide while still doing research and editing videos.
I also developed some workflows and flowcharts that could help visualize and ease the captioning process in the future and were embedded within the style guide.
Style Guide Components:
Evaluation & Assessment
Identification
Language of Origin
Linguistic elements
Prevalent Issue
Metadata Comprehension
Hemispheric Institute Video Page
Artists
Background
Video Summary
Transcripts
Konch
Tags
Mechanics
Technical Structure
Timestamps
Speakers
Editing
Contextual
Emotion
Emphasis & Enunciation
Listening
Terminology
Indigenous Languages
Music
Readings
Grammatical
Punctuation
Spelling
Italics
Review
Notes & Highlights
Typos & Grammar Check
Cohesiveness
Workflows & Other Final Infographics
Obstacles
Technical difficulties with captioning software
Sound issues
Subtitled videos within automated captioned videos
Deliverables
Complete and concise style guide
Organized and led a successful edit-a-thon event
Developed workflows to assist in creating a smooth captioning process
Concluding Thoughts
In conclusion, the style guide provides thorough methods and directions to create accessible and accurate captions that are just as an immersive experience as that of the actual performance or interview.