As legal trends and the push toward universal design increase pressure on creators of educational video to include captions for their materials, many of us are comparison shopping for caption providers. Two of the vendors you may be hearing about at Duke these days are 3Play Media and SpeakerText. James Speckart, Program Manager for Instructional Design at Duke’s Social Science Research Institute, recently conducted a detailed comparison of these two vendors, using a video he produced for SSRI. He graciously agreed to do a writeup of this work so that others at Duke can make more informed choices in their selection of a provider:
When we started to design new material for SSRI’s website, we didn’t worry about legal compliance with possible ADA legislation too much. We were trying to embrace some of the casualness of quick-and-dirty video production, and we figured that we could have some freedom from ADA compliance issues until we knew 1) what components and technical elements we wanted to have, and 2) how Duke was moving to comply with ADA rules in the coming years.
[youtube]https://youtu.be/sNpxNk083Pc[/youtube]
We knew that having transcripts for video content was a good idea just on pedagogical grounds (and a commonly requested item by users on other educational sites), but our videos try to keep an informal, unscripted style, so we knew we’d have to outsource any potential transcripts after our videos were already finished. And once we decided we’d make our own website rather than use existing resources like Sakai, we needed to figure out the technical details about transcription services, from textual accuracy to website integration tools.
We looked at the two main outside providers that Duke suggests for transcription: 3Play and SpeakerText. They have very different approaches to processing video. For $2.15/minute of video, 3Play first analyzes the video with language recognition software and then has an American transcriber correct any mistakes.
For either $.90 or $1.25/minute of video, SpeakerText doesn’t use any software but instead uses Kenyan transcribers, and offers two levels of proofreading (which explains their two price tiers). Both can be given access to a YouTube channel and upload transcripts directly to be used as captions, which is a good time-saving service. Both also offer a website plugin that provides a searchable, read-along text box containing the transcript.
So to test the two products, we sent the same YouTube video (above) to each, and in SpeakerText’s case, asked for a transcript at each level of accuracy they offer (95% and 99%). The video was one of our unscripted videos explaining a table of empirical research results, and while it had pretty good quality audio, I thought would be good tests for the transcribers because the speaker used a few technical terms and also had a few speaking mistakes sprinkled around.
The best results came from the SpeakerText 99% accuracy transcript: while 3Play also claimed 99% accuracy, it marked some words as inaudible that SpeakerText could process, and Speakertext’s extra levels of proofreading for 99% accuracy made a very useful difference. We could have simply chosen SpeakerText for financial reasons (it charges half the price of 3Play), in the end SpeakerText also produced better results. They’re not perfect, but it’s a very good product for the price and the quick turnaround time.
We decided to link to the transcript as text files for students who want to read along rather than use the read-along text box; that functionality was a little hard to fit nicely into our website design, and we’re not sure if dividing the user’s attention between text and video boxes is always a good choice. We’ll also use the transcript for YouTube’s on-screen captions. There are many other educational design issues that need to be addressed to maximize compliance with ADA issues, but we’re hoping that clear video, audio, and a transcript is a good start that will help all our users regardless of their abilities.
More information about these two companies and documents to help you get started using them can be found here: https://oit.duke.edu/voicevideoweb/video/creating/captioning.php
