Listen to Brad Boim discuss Digital Asset Management
Henrik: This is Another DAM Podcast about Digital Asset Management. I’m Henrik de Gyor. Today I’m speaking with Brad Boim. Brad, how are you?
Brad: I’m doing well, Henrik. How are you?
Henrik: Good. Brad, how are you involved with Digital Asset Management?
Brad: Well, the NFL cable network launched back in 2003 with a 24/7 linear channel and at that time, we were pretty much like a startup operation, so Digital Asset Management was not really much of a part of the conversation at that point. And I was originally hired as an Avid editor and it was a pretty small operation. We had three Avids and we were working in a pretty scaled down operation from where we are today. So early on, the amount of media files that we were accessing, storing and creating was pretty minimal. And so there really wasn’t a lot of thoughts or concerns about asset management or organization at that point. And our storage was localized to individual Avids and we were running standalone islands at that point and you know, cycling… old school cycling of videotapes between local Avid Systems. So you know, sort of did that for a few years. A couple of years into the network, we scaled out from there. So we deployed a Unity system which was a 13 terabytes Unity. So finally at that point, we had some shared storage and some workspaces across our edits systems and you know 13 terabytes probably at the time seemed like a ton of storage working at DV25. But that quickly was apparent that that was not the case. We were filling the storage up in and we realized we had to kind of address it on some level, but I don’t think that we really placed a high priority on asset management at that point.
Brad: It was really a few years later when we switched our platform and storage to a SAN environment and we flipped over to Final Cut Pro and at that point, we deployed a MAM [Media Asset Management]. So that’s where we had some tools available to us to control ingesting of game content and various shows through the MAM. And that kind of led us to a place where maybe we needed to start focusing on asset management at that point. So fast forward 15 years from day one and our post operation and asset management needs have grown exponentially. So I had shifted my role from being sort of a hybrid editor and asset manager to really moving completely into the content management side of things here. And so since that point, you know years ago, I been managing all the post-production and asset management workflows here at the NFL media group in LA.
Henrik: How does the professional American football league use Digital Asset Management?
Brad: We’re a linear cable channel and we also have multiple digital platforms that we’re supporting. So our website, nfl.com and VOD channel and mobile apps and even delivery to our league sort of controlled social media sites. So you’ve got a lot of different production groups within our facility that are looking to get access to the same media assets and do it quickly. That stuff turned around as quick as possible. So it’s pretty central for us to have the ability to index and organize all these thousands of media assets within our MAM. So we put a lot of effort sort of on the front end to get as much metadata entered into the MAM on the individual assets as possible so that people can get quick access to the content. So adding a lot of metadata tags based on our internal taxonomy system, gives producers the ability to do really quicker searches in the MAM for any specific file or play or content from a specific game or an event that they’re looking to get access to. With NFL content, you can kind of categorize it into pretty distinctive categories. So it can be game footage and all the ways of describing game footage, whether it’s season or week or play description.
Brad: You have lots of things like press conferences and studio-based content for a lot of our shows that are produced out of our facility here and plenty of other categories that people typically are looking for. So within the MAM, our media asset management system. So we’re running an asset management layer across the facility. So that is indexing the thousands and thousands of media assets. So it’s essentially a search engine for curation of all the media content. You can categorize media assets and a number of ways by team or by media type, whether it’s video content, audio or still images and there are plenty of ways of sort of providing multiple metadata tags on content so that people can really find it. So in most cases, we’re tagging the same piece of media in multiple ways. So one person might do a search based on the team that are associating the search with and then another person might do a search based on a player’s name or on a season and they’re going to get to the same media assets from either of those searches.
Brad: So the system pretty flexible and intuitive for people to find things once that metadata gets added to the content. So I think our asset management system is the centralized access point for some other metadata that we’re also trying to get in that are pretty crucial to things we do on a day-to-day. So things like, and these are things that we are really sort of developing right now, but the ability to utilize cloud-based services to do things like speech-to-text on sound press conference interviews is becoming a real-valued asset for us. And once the transcripts are received back from the cloud, we’re transferring that metadata back directly into the asset management system where it becomes time aligned with the asset so people can search off of it. So for example, if you had a Bill Belichick press conference and he spoke for 15 minutes and the producer is only interested in the time that he spoke about a certain player’s injury, instead of having to listen to the entire thing, you can search off of that keyword of injury and you would be able to narrow down the specific places where that was referenced in his press conference.
Brad: And you know, in a kind of a similar way of harvesting metadata. We’re also bringing in the play by play information for every NFL game and importing that data right into the MAM [Media Asset Management] as well. So for a typical game you’ve got every single play has a play description, that information gets time aligned onto the media asset within the MAM as a marker and then all those plays with those detailed descriptions of the play becomes searchable both in the MAM, but then when the clip gets brought into the NLE [Non-Linear Editing system], the markers come with the media assets so whether you’re a logging or editing, you can search and find pretty much any play from any game. And then a few other things that we’re using tools within our MAM that become pretty useful within the building is the ability to log in, watch proxies of clips within the MAM so that PAs or producers can spend some time prepping and adding their own user-based markers onto media and adding notes into the MAM. And then that information assists the editors because all that information transfers over to the NLE when the clip is moved from the MAM into the NLE and the NLE would be our [Non-Linear] Editing platform.
Henrik: Brad, what are the biggest challenges and successes with Digital Asset Management?
Brad: I think that the volume of media that we generate within our facility on a daily basis has really increased significantly over the past few years. So it’s exceeded the support personnel capabilities that we have in our building, which are doing a lot of manual metadata tagging within the asset management system. So this kind of wave of content makes it a little bit difficult to keep up with the pace and get all that metadata applied to the assets so that everything becomes searchable within the building. So some of the things that we have been challenged with and really would like to get to is the ability to bring a real-time stats feed of the play by play from each individual football game to get that data dumped into the MAM in real time and have it time aligned with the game footage itself. And currently, we’re doing it after the fact.
Brad: So we will get that data and we’ll import it and marry it in the asset management system to the asset after the game is over. But the real value for us would be to have that data streaming in and going associating with the media clip in realtime and that’s something that we’re looking to actively solve right now. And then I think with a lot of the machine learning and AI technology that we’ve been looking at, you know, to try to automate some of the processes that we are currently doing manually. It can be a huge help. But I think some of the tools it’s been a bit slow to try to deploy some of these things. And there’s still a little bit of a level of human interaction to be able to parse through all of this machine learning metadata that you’re generating and determine how accurate it is and what data is really going to be beneficial. And what maybe is extraneous data that you want to toss out. And I think certainly as image and facial recognition technology and the speech to text and these other machine learning tools become more accessible and integrated into our asset management system we’ll be able to figure this out a little bit better and learn how to utilize. And potentially monetize it better. There’s also always been a kind of a challenge in quantifying the benefits of, you know, spending additional time and resources into metadata curation and certainly the added expense and effort of integrating machine learning data from, from end users perspectives. I think the resources are essential for them to do their day to day jobs, you know, they all want more ways to get access to content quicker and more information for them is gonna be beneficial. But translating those benefits and the expenditures on paper to reflect those immediate cost savings benefits can be kind of complicated to push that up the chain. But I think overall, we’ve leveraged our asset management tools here much more over the past few years. We’re starting to see a lot more engagement in adding these resources from our production groups here and they are the ultimate consumers of the tools that we’re providing so it’s gratifying to see them utilize these workflows that we’re building and see that we’re creating more efficiencies for them in their day to day workflows.
Henrik: Brad, what advice would you like to share with DAM professionals and people aspiring to become DAM professionals?
Brad: Well, I think you should always be thinking that there’s a better way to do things than maybe what you’re currently doing. Our workflows and approach to organizing content in our asset management system. It is constantly is evolving and sometimes you might come up with an idea to harvest metadata, provide it to your production teams and it just kind of falls flat or they can’t utilize it effectively and from my experience, it really helps to engage with the end users of the workflows that you’re trying to develop and listen to what they’re really trying to accomplish and sometimes a great idea…it might not fit into the aspects of live TV production and the immediacy of what they’re trying to do, but it could be a more beneficial tool for your archival workflows as opposed to the quick turnaround live approach, which doesn’t always work well when you’re trying to add extra layers for them.
Brad: So in most cases, we’re trying to develop solutions for both sides of that. You know, it really does come down to how you can directly impact the efficiencies or the people that are using the system. Like I was saying before, the emerging tools like AI and machine learning, they can potentially make our lives much better by providing richer set of data that we can all utilize, but it does create a whole new set of challenges and filtering through this mountain of data and making sure you’re using it in a way where it’s being a benefit to everybody in your buildings. So those are some of the real challenges right now. The technology that has been evolving is really keeping us fluid with some of the decisions that we’re making. I think you need to maintain a lot of flexibility with your infrastructure so that you can adapt when new technologies come into play.
Henrik: Well, thanks Brad.
Brad: Yeah, thank you very much.
Henrik: For more on this, visit anotherdampodcast.com. If you have any comments or questions, please feel free to email me at email@example.com. Thanks again.
Listen to Another DAM Podcast on Amazon Alexa, Apple Podcasts, AudioBoom, CastBox, Google Play, RadioPublic, RSS or TuneIn
Pingback: 10 Things on the 10th October: DAM, Metadata, Taxonomy and Racist AI – MOD LIBRARIAN