Revinar

Revinar - Rev + Dovetail, An API Love Story



Webinar Transcription

Danny Lambert
 
Awesome, great, so I'm gonna open up my deck here, I have Kai Forsyth on my right hand side, I'm gonna be giving him a more formal introduction here in a bit, but let me just pull up the deck on my end. So as I mentioned, we have Kai Forsyth here from Dovetail, my name's Danny Lambert, I work on the Marketing Team here at Rev. And we have a really awesome customer story use case around, you know, our API usage here with Dovetail. They just got, I think, $4 million in C Funding recently, and with their beta release, they built on top of our Rev.ai API some really cool features into their market research user research software. Leveraging asynchronous transcriptions and a lot of really cool stuff that we thought we'd share with you today, get your perspective on some of the cool ways that our customers are leveraging Rev, and insight into a really cool tool that Dovetail has put together.
 
Before we go into the actual meat and potatoes, which is everything that Kai and his team have done, for those of you who may not be familiar with Rev, just wanted to give you a background on who we are, what kind of service we offer, and that will parley really nicely into Kai's portion. So as far as Rev, we're a provider of transcription, caption, and subtitle services. We have over 100,000 customers, and that spans everything from global enterprises like Facebook to your local favorite YouTube influencer and everything in-between. Rev processes more audio and video than any other provider in the space, and we do that using a network of over 50,000 transcriptionists, primarily based in the U.S. We've transcribed more than 100 million minutes for companies like Google, Facebook, Buzzfeed, and the like. As far as who we serve, 60% of the Fortune 500, 25% of the Global 2.00, and everyone in-between. As you can see, there's kind of a smorgasbord of logos back here, but there's pretty much no company too large or too small who has any sort of speech-to-text needs. If you're producing audio or video, and you need any of those services that I mentioned before, no one too big or too small to use Rev.
 
As I mentioned in the beginning, I said APIs, and you're darned right that we did. Most of our customers are super familiar with the self-service way to purchase through us, which is going through our website, you upload a video, you get transcriptions or captions back, or maybe use us through YouTube, or a content management system like Panopto or Kaltura. But we also have a whole other side, a developer side, to our business, where you can build on top of the same engines that we use to provide captions, transcriptions, to our customers. So this is a nice little graphic here, we have a Rev.com API, and an Rev.ai API. Dovetail, specifically, is currently using the Rev.ai API, but there's minor nuances between which one you would use, and for what services. So if you're looking for human transcription, human captions, so you need really high accuracy for ADA compliance, the human route is typically the better way to go. You get a human verified, typically in less than 24 hours turnaround time, file back through that API, which is not available from Rev.ai.
 
On the flip side of that, the Rev.ai API allows you to do synchronous transcription caption. So streaming, so if you have live broadcast media, or if you're looking to have anything transcribed or captioned in real time on your files, that's only available through the Rev.ai API. And in the in-between, there's that asynchronous, automatic, transcription captions, you could technically use either of the APIs for this, there's just minor nuances on like how that's delivered, and pricing. The Rev.com API is more user friendly. You get like an editor, the pricing is a little bit higher for some of those add-on features, whereas asynchronous transcription captions through Rev.ai is more like developer focused. To add a little bit of color onto like how people are using this, before we hop into specifically how Dovetail is using this, on the human side, imagine that you have a huge CMS system, or a huge content management library that you want to get transcribed by humans and have ADA complied that you can distribute to your university, or on your website. That would be a really good use case for human transcription.
 
If you want to white label our captions or transcript editor, which we'll go into a little today, or if you had transcription scale that was really large, imagine a call center, or a digital asset management software, the asynchronous, you know, somewhat near real time transcription, at a high volume, is typically what people are looking for. And then from a streaming perspective, you see this a lot with live captions of in-person events, or because they had own dog food and built a Zoom live-captions integration that's coming out here in June so that whenever you're on Zoom conference, you can get live captions for all those live-streaming needs. And as I mentioned, it was using Rev, who's using Rev.ai?
 
Typically it's our engineers, you know, pioneering top brands, innovating with speech detect, so anyone who is using a lot of audio or video, specifically who's making that more accessible, to people online. These are some of the logos that are currently leveraging either our Rev or our Rev.ai API. So without further ado, I'd love to introduce Kai Forsyth. He's the Product Marketer at Dovetail. Dovetail helps you store, analyze, and collaborate on user research in one place, making it easier to see patterns, discover customer insights, and decide what to do next. Prior to joining Dovetail, Kai worked on research and design projects at Intercom and Atlassian. Kai, thank you so much for joining us today. It's a real pleasure to have you on.
 
Kai Forsyth
 
No worries, yeah, it's good to be here. I'm super excited to share what we've built and how we've integrated with Rev.
 
Danny Lambert
 
Awesome, man. If you guys aren't familiar with Dovetail, Kai's been nice enough to kind of pre-record a live product demo. We know most of us are remote, so we wanted to make sure that everything went off without a hitch. But he's going to hop right into what his team's been able to build and show you first hand how awesome that is. Then we'll go into a Q&A session with him after this.
 
Kai Forsyth
 
Good shot. So yeah, effectively that customers are doing in Dovetail. The tour really helps you go from unstructured text and qualitative content and gives some structure to that through tagging to find patterns across your data. So practically what this might mean is if you're doing research, if you're a product manager, or if you're a designer or researcher, you're doing a lot of interviews, a lot of customer interviews, you'll want to transcribe those and find the patterns across the text. So there's a clear-use case for transcription and integration with Rev, here. So I'll show you what we built and how that practically works in Dovetail.
 
So this is our interface. We put a board view here. We're going to upload a recording from maybe a Zoom call, or a Google Hangouts call into Dovetail. So this is about four minutes. While that upload starts, we'll title this interview here. And then while it's uploading, we're going to request a transcription. So when that file finishes uploading, we split out the audio from the video, and then we send that off to Rev, and then the transcript comes back. So while that's processing, we'll jump into some sample data. So here we've got an interview with Davina, and the transcript has already come back of the video there.
 
So this is the integration we built. So you can click play and once it loads, it's going to do... It's going to speak the word that's being spoken. So Rev's API returns the timestamp for each word. Which allows us to build this integration where you can sync the spoken word with the video playback. You can see that there. We got the speaker names back from Rev. We built the ability to split the monologue, fully edit the transcript in our custom editor. We can jump to sync and jump around the video based on where you click. And this is the bread and butter of the pill here, as I mentioned, creating those tags in the transcript content to find patterns across the interviews.
 
So you can see we're creating one here, the Working with People. So this is showing off renaming speakers, so if we want to change this to Paul, we can do that. It's going to change all the mentions. Or we can override any speakers we get back, and add Susie for this speaker here. So I think we're going to jump through to the Working with People tag. So on each tag you can see the snippets of each video that has been tagged with that content. So here we've got an interview with Davina, with Lucy, and we've built the ability to play it, and it'll only play that section of the video based on the text that you've selected, which we're getting from the time stamp that comes back from the Rev transcript through their API.
 
So you can see here, there's a 30 second clip, which is the, sorry a 17 second clip of the 17 seconds of that sentence we selected being spoken. So you can see, across all the interviews, with all of the mentions of Working with People. So project managers and researchers can understand a bit of the patterns across their data set. And you can click through to any interview like this one with Sherif, and be taken to the specific part in the transcript, keeping that context, with those tags.
 
Danny Lambert
 
That's really cool to see Kai. I don't know if any of you on the webinar do a lot of user research, a lot of market research, but we've been doing a bunch in the development, recently, of Zoom captions, as well as some specialty stuff for Edu, and managing this out of Google Sheets, or AirTable, or wherever else we had. We would have killed to have a feature like this available at the time. So really, really powerful stuff to see what you guys are able to do.
 
Kai Forsyth
 
Yeah. So I think that transcript's come back now, you can see it's taken the first frame of the video, you can see my face there in all of them. But the upload and transcript that we requested at the start has come back, and you can see this is the experience as it comes back from Rev. So you can click play, and visit the raw transcript that's been returned. We can name any speakers to the actual people. We can override them, again, the playback experience just works out of the box. And if there's anything you want to tidy up with the transcript, if you want to split the monologues out, make it a bit more readable, you can totally do that. Delete some of the disfluencies that might have come through. Yeah. That's basically it, in a nutshell. Again, creating the tagging experience. You can see the whole transcript there.
 
Danny Lambert
 
That's very cool. Awesome. Is that the end of your video? I don't want to cut you off. That's really cool, Kai. I was playing around in your beta a few weeks ago, and just testing all these features, and I was just like, "Man, if only I had this months ago, how easy my life would have been." Some of the initial questions that came to my mind, you know, as I learn more about your process, more about what you develop, was understanding how you heard about Rev.ai, and what it was like when you were vetting different vendors. I'm sure there's other softwares out there that you tested to do the same thing. How'd you hear about us, and what were you looking for in a vendor?
 
Kai Forsyth
 
For sure. Absolutely. I think, we were definitely aware of Rev, to start with, a lot of our customers in the research space, customers of Rev, the human product, and requests for integrated transcription was something that was coming directly from our customers. All of them were, it seemed like they were using Rev. They always mentioned Rev in the feature request. So, we wanted transcription, but we wanted to get further and really integrate it with the video and really tie that experience together.
 
So we were looking for a vendor that would really allow us to deliver that capability. And I think the way that Rev's transcription API returns the transcript content really allowed us to sync each specific word to the video playback and really deliver this experience that is so much richer than just having the transcript text back. So I think that was one main requirement. And then accuracy is super important, too. So there's really this trade off, I think, with the amount of time that it's going to take to get an accurate transcript, and the speed, and whether there's tolerance for errors. I think typically a lot of researchers in organizations have a bias towards speed and getting something back that they can work with rather than something that's particularly accurate. So, you know, we wanted as much accuracy as we could with the machine-generated transcript, and Rev, in our testing, was the leader there for the most part.
 
Danny Lambert
 
That's awesome. How about the whole developer experience for you guys? Like once you found out that Rev.ai was a possible API you could use for this, how was it working with the docks, how was it developing our API. Could you shed some light on the developer experience?
 
Kai Forsyth
 
For sure. I think it was pretty easy, to be honest. I think the engineering team really put together a prototype, like a proof of concept quite quickly. We were able to send content to Rev and get a transcript back. I think the challenging part for us was really the stuff outside of the transcript, and outside of Rev. Around the video processing and integrating the video into our editor, and all of that, I think that's a lot of the engineering effort went. But in terms of the Rev API and the docks, pretty straight forward. Just able to get up and go in. Our engineering team had no issues at all there.
 
Danny Lambert
 
That's real awesome to hear. And I know the beta only went live like a week and a half, two weeks ago. What's kind of the reception been like, from your users? How did that launch go for you guys?
 
Kai Forsyth
 
Yeah, really good. I think the feedback has been incredible, actually. I think a lot of researchers, they would be pasting in transcripts from Rev into Dovetail, you know, transcribing things themselves, so to be able to have that and the video linked, and just upload a file from an interview, it seems to be really positively received. I mean we've processed a pretty significant volume of video and transcripts back, and yeah, it's been incredible. What's interesting is that before transcripts, we had... You would just be creating tags in text. But we're now seeing that tags that are created on transcripts represent kind of half of all the tags that have been created in Dovetail.
 
Danny Lambert
 
Wow, that's quite the shift.
 
Kai Forsyth
 
Yeah, absolutely.
 
Danny Lambert
 
I'd also be interested to hear, and I know we kind of talked offline about this, you, and along the lines of, who would use the human API, why would you leverage that, verus maybe the automated APIs, we do have them both. I know this is something that your team has been talking through and working through. Do you mind shedding some light, from your seat, as to why you chose to start with Rev.ai, what the trade offs are between human and automated?
 
Kai Forsyth
 
For sure, yeah, it's a good question. I think there's a few that play into it. As I was mentioning before, it's like the speed and how quickly you need to get something back that you can work with, the machine-generated stuff is just by nature... Machine-generated is going to be quicker. The other concern that's quite important to call out, a lot of our customers, they're entering personal conversations as part of their research. And a lot of them feel a lot more comfortable, I think, having that processed by a machine, by an API, rather than a real human. So that's another thing to kind of consider, I think. Yeah.
 
Danny Lambert
 
Absolutely. Yeah, it's... It's one that when you're recommending to someone, it's like you'll know it when you see it a lot of the times. You either need a certain accuracy, or a certain speed, and it's cool to see that it's part for you guys how you made that decision today. Something that I always get questions around, and especially with us being relatively small, small team, is how has it been with working with our development team and support? If you guys have hit any snags? What's that process been like? Giving product feedback, getting responses from our developer support?
 
Kai Forsyth
 
Yeah, it's been pretty good. I don't think we've really needed to that much, but when we have, it's a matter of reaching out, shooting an email, getting a result. It's been pretty easy, honestly. I think as part of, the feature was quite involved, to build, but the transcription part of it was probably one of the easiest parts. The video processing and all of that, and the streaming, and the playback, was kind of new territory for our engineering team. And I think that's where the challenges were. But working with the transcript, then getting the response back from Rev, was relatively straight forward.
 
Danny Lambert
 
That's awesome. So for anyone who's on right now, if you want to ask any questions of Kai, we do have him for a little bit longer, so please feel free if you have any questions, you can drop them in the questions tab, or you can put them into the chat. While you guys think about those questions, I do want to show you guys an offer where you can, either try the Rev API... Can you see my screen? All right, if you'd like to try to Rev.ai API, you can get five hours of credit for free, so if you go to Enterprise.Rev.com/Rev.ai, you can get started. Use exactly the same API that Kai and his team are. Test this out with whatever your workflow is. Also, if you're doing any sort of market research, or if you any departments in your organization that are doing user and market research, please go over to Dovetailapp.com, give this a try, they have I think it's a free seven day trial, you can test all the features that we covered today, and you can give this a try for your team, let them see if this will have value.
 
I know that we're already going to be transitioning a lot of our market research into here just so we can tag it and get like stackable, easily findable key sentences and information from all the customer research that we do. So please, don't hesitate to visit either of these links, or try out both of these. And then, as an aside, if you ever want to reach out to me, it's danny@rev.com, and Kai is also nice enough to include his information. So if you have any questions about the product, about the process, or just want to connect with him, his email is on here as well. Kai@dovetailapp.com We'll be sending the deck out afterwards, so if you didn't have a chance to jot this down, no worries, it'll come through in a follow-up email either later today or early tomorrow.
 
And on that note, I know I saw one question come in. Let's see if we can't answer that for Fran. So Fran's question is... Can multiple people be tagging the same interview simultaneously? The example says three grad students who collaborate on tagging at the same time to tag video interviews faster.
 
Kai Forsyth
 
Absolutely. So we had our editor built, before the transcription integration, the editor is a fully editable, collaborative, editor that we built. And through the integration, we just added the transcripts into that editor. So we've got all of the collaboration features for free, so you can have multiple people in the document, editing simultaneously, creating tags simultaneously. It works pretty well, up until a limit, I guess. You probably don't want ten people jumping in and editing the same document. But yeah.
 
Danny Lambert
 
Awesome. Thank you Kai, thank you Fran for that question. I think your question as well, Mary. Can you help us understand your fees? Is that in reference to Dovetail's fees in using their product, or the fees for using the API? I believe that one's for you, Kai.
 
Kai Forsyth
 
For Dovetail? Yeah, so our platform, there's kind of a broad range of features on our platform. Today we just really dived into transcription. We start at $125.00 a month, and that includes five contributors. And transcription, at the moment, and the video functionality, that's free while in beta. There will be some price factor to that, which we're still kind of working out. There will also be some minutes included in the plan by default. But that's, we're still kind of working through that. But .
 
Danny Lambert
 
Awesome. And then, Mary, as far as the Rev side, it depends on what service you're using. The human's going to be a little bit more expensive than the automated. And if you have used Rev, the pricing is a little bit different. If you go through the self-serve, you're used to paying, you know, $1.25 a minute. The API is based on a per second basis. So a little bit different than what you're probably used to if you've worked through Rev.com before. I don't have the exact numbers, on the top of my head, but I do have your email, so after we're done with the call, Mary, I will shoot you the rates by the different APIs and what the cost per second is. Awesome, well, it doesn't seem like we have any other questions. If we have any come in after the fact, I'll just edit this video and then update it to answer any other questions that come in afterwards, or come in via email.
 
After this is all over, you'll receive a video recording of this, so if you'd like to share this with anyone on your team, or anyone who might be interested in either Dovetail for their market research, user research, or for leveraging Rev's APIs, please feel free to share it. It'll always be available after the call. And thank you again so much, Kai, for joining. This is really cool to see. I always love seeing the awesome ways that people are leveraging our APIs, and your product has been really fun for me to use and hopefully for the audience as well.
 
Kai Forsyth
 
Sure, yeah, it's been great. Thanks for having me, and plugging through it. Really glad to be partnering with Rev on this. Yeah.
 
Danny Lambert
 
Awesome. Thanks a lot, Kai. It's also very early, it's 6:00 a.m. for him there in Australia, so I really appreciate you hopping on for us. Awesome. You have a great day.