Best Practices for Automatic Transcript Alignment
JOSH MILLER: Hey everyone. Welcome to today’s webinar. My name is Josh Miller. I’m one of the co-founders of 3PlayMedia. Thank you for joining us.
So today, we have two experts on this transcript alignment service that we recently launched. We have Roger Zimmerman, who is our VP of research and development here at 3Play, and David Zylber, who is our manager of customer happiness. So the two of them are going to go through a number of great details on the alignment service. So Dave, why don’t you go ahead and get us started by telling us what we’re going to cover.
DAVID ZYLBER: Thanks Josh. Today, we’re going to cover everything you need to know about the transcript alignment service. We’re covering submitting transcripts and media files, formatting your transcripts, and most importantly, the best practices to adhere to when submitting your transcripts for alignment. We’ll also explore how this automated process works. For anyone who’s familiar with the 3Play account system, this all should look fairly straightforward.
JOSH MILLER: Great. So why don’t we start by explaining exactly what this transcript alignment service is and how it’s different from the standard captioning and transcription service that a number of people are used to seeing from us.
Transcript Alignment vs. Standard Transcription & Captioning
DAVID ZYLBER: Sure. Well, the alignment service is a great option for our customers who already have transcripts for their media content, whereas you’d want to use the standard transcription and captioning service if you didn’t already have a transcript for your video. The transcript alignment service offers 3Play account holders a faster and less expensive way to create and use our interactive plug-ins by synchronizing your non-3Play created transcript with your media files. This differs from the default 3Play Media service transcription and captioning, where we create the core text document from either a video or audio file that is uploaded to the system.
And then once your alignment is complete, you’ll have access to all the products and services that come with our standard transcription and captioning service. This means you’ll be able to download captions and transcripts in a variety of formats, create translations and multilingual subtitles, and use the interactive transcript and captions plug-in. You’ll also be able to use the Clipmaker video search and clipping tool, captions and subtitles editors, and any other features, basically, that you would normally have access to.
JOSH MILLER: Great. So in terms of process, Roger, can you explain what some of the major differences are between this alignment service and the transcription and captioning that we normally do?
What Makes the Transcript Alignment Process Different?
ROGER ZIMMERMAN: Sure Josh. Hi everybody. This is Roger. I’m happy to be here to discuss this exciting service and the technology behind it.
A major difference between the default transcription and captioning service and the transcript alignment service is that the transcription and captioning service combines an automated process, followed by human cleanup by our team of expert transcription editors and QA specialists. The alignment service, on the other hand, is essentially 100% automated. And that’s very important to remember as we go through this webinar.
So just to elaborate a little bit more on those differences. So when using the standard service– And those of you who have used that service will understand that, we, of course, have no transcript. So the media file that you upload initially goes through ASR, which is the acronym for automatic speech recognition, to create a transcript that is about 70% to 80% accurate.
Now, as everybody can probably understand, that is not good enough for most realistic applications. So 3Play’s transcription and captioning service uses these human editors and QA specialists to ensure a nearly 100% accurate transcript. And again, the ASR output goes through two rounds of human cleanup in order to get to that accuracy.
The alignment process, on the other hand, is 100% automated. Doesn’t go through any cleanup service. That means that when you submit your transcript to be aligned with a media file, it needs to be formatted and its content needs to be massaged, if you will, if you want to achieve the best results. And we’ll get back to that shortly.
JOSH MILLER: Great. All right. So, how about the technology itself? Can we walk through what is actually happening in the background?
Understanding the Technology Behind Transcript Alignment
ROGER ZIMMERMAN: Sure. So, the first step in this technology is actually one in which the customer interacts directly with the technology. And this is where you are uploading your transcript to the system and associating it with the appropriate media file. Dave will go into some of the mechanics of that later.
But the key concept here is that the automated system requires that the transcript be entirely in pure ASCII text. Now, some of the transcripts that you have will already be in ASCIIs, but others may not. And in particular, if your transcript document was originally created in MS-Word, you will need to export it to text format prior to uploading to our system.
Now, you should recognize, however, that Word does not use standard US ASCII format. It uses an extended form of ASCII. And as such, the technology needs to convert any non-ASCII character produced by Word to the nearest ASCII equivalent. If it succeeds, you will see the text in the interface looks normal, with quotes and dashes successfully converted.
If however it does not succeed, you will see question marks in the interface. And this may occur for transcripts produced in other word processing programs or by older versions of Microsoft Word. If this occurs, you will need to either manually edit your transcript or work with our support department to get your document into the right format. So Dave, why don’t you take it from there?