Transcription Today: An Interview With Larry Jordan
The following transcript is taken from the Digital Production Buzz podcast released on the 25th April. To listen to Dom’s interview with Larry Jordan click here.
Larry Jordan: Dom Bourne is the founder and president of Take 1. This is a UK based broadcast metadata and transcription service. But he’s also a keen technologist. He’s responsible for the company’s strategic partnerships, research and development and future proofing. Hello Dom, welcome back.
Dom Bourne: Thank you Larry, it’s good to be on the show.
Larry Jordan: Dom, as head of an international transcription firm, I want to talk with you today about the future of transcription, but before we do, give us a brief description of what Take 1 does.
Dom Bourne: Take 1 processes huge quantities of video files for its customers. They might be video production companies, they might be networks, they might be localization partners. And we have a head office in the UK through which all our servers are based and we receive video files from all over the globe, and we process those using a combination of technology but also good old fashioned humans to get the quality standards that our customers demand.
Larry Jordan: I’ve had the pleasure of working with you and your team for many years, five or six years at least. How has the process of creating transcripts evolved over the last five years?
Dom Bourne: Transcripts have evolved over the last five years in a number of interesting ways. Our customers are wanting to integrate transcription services tighter with their editing workflows, so we’re seeing a lot of customers wanting ScriptSync compatible transcripts for example, whereas five years ago, they would be very interested in just receiving a Word file. We’ve also noticed that customers want faster and of course cheaper transcripts, but that doesn’t always sit comfortably when they also want accuracy as well. The customer ideally wants fast, cheap and accurate but in the real world, you can’t always deliver all three in the way that the customer wants. If they want it superfast, they may have to accept the accuracy won’t be as high.
Larry Jordan: One of the things that you taught me, as we were chatting over the years, is that what customers want from transcripts has evolved. It used to be as you said, just a straight Word doc, but now they need closed captions, they need multiple languages. What all are you being asked to do these days?
Dom Bourne: For the most part transcripts are still being used as a curation method to distill high shooting ratios if you’re making a reality show. So that’s really an editing workflow use case, but there are a number of other use cases such as if you’re a video publisher, wanting to get your content out on the web, then you’re going to be interested in having captions to go along with your video to make sure it can be reached by wider audiences. We also have a different camp of customer who is interested in localizing their TV shows and for those types of clients, what they’re really interested in is a highly accurate blueprint which we can deliver, sometimes in XML or sometimes just as a text file, but either way it’s the cornerstone of the entire localization process and you can imagine they need accuracy up front because if they don’t get it, then by the time they’ve localized into 50 languages, they’ve got 50 problems.
Larry Jordan: We were talking last week with Jim Tierney who is the CEO of Digital Anarchy and has an automated transcription service called Transcriptive. Digital Heaven in the UK has got SpeedScriber and I can throw a rock and hit five or six others, which are using AI and machine language to be able to create transcripts automatically. What’s been the impact of this on your traditional transcription service?
Dom Bourne: I think ASR is making inroads into all sorts of areas in post-production and transcription is certainly one of those. A lot of these so called round trip services rely heavily on the same ASR engine, to deliver what you get back. One man band filmmakers and video producers who aren’t so bothered about high accuracy, these services are great because you can hit a button and you send your video up to a service in the cloud, and you get back a transcript very quickly.
Dom Bourne: We have evaluated a lot of these engines, and whilst they’re improving over time, they’re not always delivering the product that our clients need, for example they don’t necessarily come back with accurate timecodes or accurate speaker labels. And for some clients that’s fine, because all they want to do is just do a keyword search, all they want to do is get a sense of what was discussed on an interview. They’re not that interested in a full transcript. But for sure, ASR and AI is everywhere now. It’s on our phones, it’s in our TVs, it’s in our NLE plugins, and I think it’s a very logical way of using that technology.
Larry Jordan: One of the comments Jim made which made me think of you is he said when the accuracy of automated speech transcription gets below 90 percent, it takes you longer to correct it than to do it right the first time. So when should people consider using automated transcripts, and when does human based transcription rise to the front?
Dom Bourne: Even with the automated systems, somebody’s got to do the cleanup. You can do it yourself if the vendor that you’re using has an interface for facilitating the cleanup process. But you’ve still got to sit there and do it. So it really comes down to personal choice. If you’re an individual who likes to hit the delete key and get a headache from going through a poorly transcribed interview and making it a polished interview, then be my guest. But if you want to send a file to Take 1 because ASR isn’t delivering the quality that you need, then there is no substitute for a professional service and I think there’s a reason why Take 1 is the go to transcriber of choice for The Buzz.
Larry Jordan: It is indeed, and by the way I want to thank you and your team for transcribing The Buzz each week. We are always grateful for your help and that help goes back many years, so thank you for that.
Dom Bourne: Not at all.
Larry Jordan: What do you see as the future of transcription, because it sounds like you’re painting a picture where automatic speech recognition is fast, but not accurate? And human based transcription is less fast but much more accurate and much more flexible. Have I painted the right picture?
Dom Bourne: I think you have, but as I said, ASR is improving all the time, and there are pockets of our organization where we deploy ASR where we see that it delivers operational efficiency. But you asked me about the future and what we’re seeing is an increasing number of workflows that use accurate transcripts as their starting point. For example, editing workflows, access services, localization, and compliance workflows just to name a few.
Dom Bourne: At Take 1 what we’re seeing is greater volumes of content being produced, and edited, and that relies heavily on transcription workflows. We’re seeing an increasing number of clients embracing cloud services and needing secure vendors. So for me, what I think the future holds is probably a blend of leveraging the best of breed ASR engines and technology with accurate eardrums and eyeballs who are capable of delivering a high grade product that really underpins a lot of those aforementioned workflows.
Larry Jordan: Are you thinking about migrating anything to the cloud?
Dom Bourne: Yes for sure, in fact that’s a very timely question because we have just launched the Take 1 Cloud which is our new transcription platform which is a customer facing platform. It’s already got some customers in Los Angeles which we’re very excited about, and the reason why it’s getting traction is because it’s a very secure place through which customers can upload their content and select their turnaround times for transcription. What we’re noticing is with a lot of our larger customers, particularly ones who are owned by big film studios, who have much tighter security requirements, this cloud system that we’ve launched, the Take 1 Cloud, is really hitting that sweet spot for them which allows files to be processed securely, but also effectively and easily through a nice, intuitive interface.
Larry Jordan: So what is it that gets you out of bed in a morning? What’s got you excited about the industry?
Dom Bourne: I go to the trade shows like everyone else. And I just love seeing how the new technology is being deployed in new and exciting ways. Whether that’s ASR transcription to speed up editing or whether that’s the latest camera codecs to compress files for faster moving around the planet. I just love all of this stuff and whenever I come across a good vendor, the first thing I do is tell my friends about it and Tweet about it. I have to say it was a real shame that I didn’t get out to NAB this year but I caught up with some of your interviews that you held on the show floor and they really filled in some blanks for me.
Larry Jordan: Thank you. For people that want more information about the services that Take 1 makes available, where can they go on the web?
Dom Bourne: Our website is www.take1.tv and all our services and products are available right there on the website.
Larry Jordan: That website is all one word, take1.tv, not .com, take1.tv and Dom Bourne is the founder and president of Take 1, and Dom, thanks for joining us today.
Dom Bourne: Thank you Larry.
To listen to the full Digital Production Buzz Podcast click here