SPARK Transcriber Explained

Summary: Sparkhound's newest solution, the SPARK Transcriber, helps you save time and money on transcribing interviews. This service is a part of Microsoft's Azure, and can be implemented into your system in less than a day.   

Alex:

All right, Dave, so we're here today to talk about our new SPARK transcriber solution that Sparkhound has put together. And this came up, I know because of some needs that some of your customers were having. So, tell us a little bit about how we got here, what led to a need for the SPARK transcriber?

Dave: 

Yeah, sure. There's a lot of my public sector agencies out there that do things like interviews. They record hearings and so forth. And part of the process of that, after the interview is done or after the hearing is completed, is they have to send that off to get that audio or video file transcribed into a format that they can then use it in documentation.

That process can take a while. It's a hour long process to do the recording and then it may take two or three hours to actually transcribe that recording and get it back from the transcription service. Or if you do it in house, it takes someone a while to do that and it takes  their time up. It's a very cumbersome and tedious process.

With the inception of AI these days and the capabilities the tools have there's got to be a better way to do this. So we had some companies come to us and Microsoft and ask, "Is there a better way to skin this cat?" And of course, the answer is yes.

Alex:

So, tell us about the SPARK transcriber? What did we put together?

Dave:

So there are a lot of transcription services available through Azure these days. You can simply upload a file; it will read that file, and transcribe it out into a document that you can publish. You can do diarisation with it. It can tell you if there's multiple speakers in there. You can assign how many speakers are in the room and put names to the speakers. It'll give you a confidence level of how well it thinks that the speaking part is being properly transcribed by the solution.

So it could be like 87% confidence level, or a 27% confidence level. It gives you an idea of how well it's doing the translation and then give you the option to go in there and say, "OK, if this didn't quite read write, let me go back and listen to it." Now you have the ability to edit just that one section and clarify what the terminology is being used at that time.

Alex:

So it sounds like a lot of stuff here, but I know that we're getting this set up rather quickly, right? This is not that lengthy of a project at all, right?

Dave: 

Right. The 1st iteration we did of this took us a couple of days to get everything in place. The tools, the transcription tools and services are already in Azure. You just have to have the licenses and basically turn it on. It's a very cheap service to use. All you really need to do is enable those services.

We made a nice pretty little interface for it where you can log in, you can see the upload, you can see the screen that shows you everything you've transcribed. You know, you can put a name to it. If you want to add more metadata like the date, who the parties are in the conversation and so forth, you can add as many metadata fields as you want to try and keep track of things.

Now that it's been created, it really takes us about an hour or so to get it installed into the client's environment. And that really just depends on if they have the Azure environment. Are they stood up? Do they have the M365 environment? And then what do we have to do? Do we want to, do we want to link it in with their Azure Active Directory for authentication? Do they just want to use the forms based authentication that we built into the system? And then where do they want to store the files? Do they want to just download them locally and put them somewhere where they're keeping them or do they want to integrate it with something like their SharePoint environment? Now they've got this library of transcripts that's searchable and readable by the intranet solution.

Alex:

So a couple of different options which determine the length of time it takes to actually implement the solution once they've decided they'd like it.

Dave:

Yeah, and it's so easy to use, which is important because you've got non-technical people typically using these sort of things and easy to find the stuff at the end, right. Easy to sort it.

Alex:

So, so tell us then with the organizations we've done this for what, what sort of impact has it has it had so far?

Dave: 

Well, [in one case, we worked with a company that was experiencing an extreme time loss to these types of services. One hour interviews were] taking them a couple of weeks to go from interview to actual transcription. That same one hour interview was uploaded into the SPARK Transcriber and was done processing in 5 to 10 minutes. They could see the output right there on the screen and they could export it out to the Word document and do any kind of editing that they wanted. So it's saved them a tremendous amount of time.

And in some cases, they're using a transcriber service, like a court reporting service to do this for them. The amount of money they were spending on that court reporting service is now being saved almost 100% because the solution only costs pennies to transcribe, where it was costing them upwards of a couple of $100 to do a transcription of an hour long interview with a suspect.

Not to mention the backlog. They could take 10 interviews and process them all at the same time. You upload, it starts processing. So it, it's not like you have to just wait for one to finish before you start the next one. You want to upload 10 and have 10 be done at the same time. No problem. What would take them days, if not weeks to accomplish can now be done in essentially minutes.

And this is definitely a scenario when organizations are asking themselves and asking us how do I get started with AI? What's a practical way to do that? This is leveraging Azure Cognitive Services, the open AI platform to do just that in a very non-technical manner. It's plug and play and you're instantly seeing ROI and some of the time saved. And the solution is extensible as well.

So if you want to take in the ability to then summarize that conversation like you can do with Teams premium with your meeting recap, if you want to add that functionality to the solution, you could easily take that 60-70 page transcript and request a summary. We can add that that functionality into the solution.

Then you can put that somewhere in SharePoint or some other solution where you can then mine that information. So if someone comes back and says, "Gosh, you know, we did an interview with John Doe, Does anyone know where that transcript might be?" Well, if you're just saving it locally on your hard drive or you're putting it in a file share somewhere, it's not really accessible at that point. But if you put it into a platform like SharePoint, well, now that's accessible to the tools search for you, and it gives you a whole lot more insight into what you have.

Alex:

Very cool. Well, exciting stuff. Thank you, Dave, for your time today. Look forward to seeing again.

To learn more about this service, or to see how easily we can implement it into your Microsoft environment, contact us.