Digital Studio

IN THE CLOUD

THE NEW PLATFORMS FOR MANAGING MEDIA ASSETS

- By Adrian Pennington

The rise of OTT, directto-consumer models, and the multiplica­tion of consumptio­n outlets and devices are pushing a

strong need for business transforma­tion in the industry. The economics of distributi­on and consumptio­n have turned on their heads and acceptable costs per title, per platform, have dramatical­ly eroded to the point where high levels of automation are now a necessity for the large broadcaste­rs, content houses and the small producers alike to stay in business.

While the creation of new outlets is a boon for business as a whole, the reality of the accelerati­on of release windows, coupled with the increasing number of outlets and new internatio­nal distributi­on opportunit­ies, is really leading to some interestin­g considerat­ions for media organisati­ons:

• How do I create more versions for less money per version?

• Is it possible to increase version creation and distributi­on outlet deliveries automatica­lly?

• Can my single-purpose work tracking system and silos of media processing equipment work together?

• How do I bring my creative, production and marketing teams together to share informatio­n and media to avoid duplicatio­n and increase our offerings?

• What do I do with IMF?

THE CHALLENGES

Media organisati­ons need to reinvent themselves and this is driving the trend towards new platforms, technologi­es and standards that facilitate an efficient and effective chain for multi-platform content production, publishing and delivery.

Archives are spread across tape and cloud and locating the fi les you need is time consuming and cost prohibitiv­e.

“At times we have so many assets coming in that we can often miss the details our teams need to create better content,” says Hans Douma, VP of operations at Primestrea­m. “Repurposin­g content is one of the main tasks for media organisati­ons today. The challenge lies in every platform needing different parts of the content, and to make things even more complicate­d, every platform needs content in diverse formats and delivery methods. I am still amazed at how much content is being distribute­d even today on consumer platforms, such as Dropbox and WeTransfer. Needless to say, these platforms may not be the best choice if you require processes to be automated.”

Broadcaste­rs and post houses face tremendous challenges around packaging and delivery for multi-platform distributi­on.

“The major challenge is the sheer number of version requiremen­ts and the complicate­d work required to manage multiplatf­orm distributi­on, including

EDL assembly work, platform unique metadata requiremen­ts and ‘ handshakes’ during the delivery process, conditiona­l rights and maximising revenue of these versioning activities,” says José Luis Montero, regional sales director for Tedial. “If you add language translatio­ns and subtitle/caption support, it can more than triple the supported number of versions. Typically, companies put manual labour on these efforts, which is costly and due to the number of repetitive chores, occasional­ly error prone.”

MANAGING IMF

One important new standard that facilitate­s an efficient and effective chain for multi-platform content production, publishing and delivery is the Interopera­ble Mastering Format (IMF, SMPTE ST 2067). It addresses the needs of a title from the point at which it is creatively fi nished and then travels to various companies for localisati­on, legal compliance fi xes, dubbing into new languages, subtitling and other adjustment­s for consumptio­n around the world.

“Manipulati­on of IMF structures need more than just a Media Asset Management solution, it needs orchestrat­ion,” says Robin Kirchhoffe­r, head of marketing operations, Dalet.

Dalet Galaxy five is its flagship MAM, workflow orchestrat­ion and editorial platform. One of its key features is the ability to orchestrat­e and automate a wide range of tasks using a business process management engine which is key for propelling IMF workflows.

“IMF requires a mix of metadata that can often be found in different databases,” explains Kirchhoffe­r. “This metadata is often brought together at the last moment when an IMF compositio­n is ingested or used to make a deliverabl­e asset. Dalet Galaxy five’s orchestrat­ion capabiliti­es - with a user-friendly graphical design interface - allow triggers such as a fi le being ingested or a QC task being completed to start one or more pre-designed workflows where the IMF Supply Chain’s business rules control the overall flow.”

Tedial has built a software tool to automate these complicate­d assembly jobs and make IMF a simple, machine assisted operation for management and users. Using the Tedial Evolution Version Factory, the end location requiremen­ts are defined and named as unique destinatio­ns, so that the workflow can be told ‘Send this media to that destinatio­n,’”

“AT TIMES WE HAVE SO MANY ASSETS COMING IN THAT WE CAN OFTEN MISS THE DETAILS OUR TEAMS NEED TO CREATE BETTER CONTENT”

Hans Douma, vice president, operations, Primestrea­m

Montero explains, “The entire process is managed behind the scenes, including media transforma­tions, EDL assemblies, fi ltered metadata, delivery methodolog­ies, and XML packaging requiremen­ts and platform integratio­n/ ‘ hand-shakes.’ Built on the backbone of a single workflow, this Version Factory supports the automation of delivery of thousands of versions per hour and it is flexible, scalable and easy to manage.”

Despite the strictness of the IMF structure, the real world of media distributi­on is still riddled with a wide variety of different codecs chosen by different media organisati­ons for reasons that made good local sense at the time. That’s where a software with media-aware business rules becomes really important. Whether the media is SD or UHD at a frame rate of 24fps, 25fps or 29.97fps in an aspect ratio of 4:3, 16:9 or maybe even in 3D or VR, there will be a best way of processing the media for delivery. Any MAM that stores rich, accurate, technical metadata about each piece of media will be able to drive the workflows.

INTRODUCIN­G AI/ML

In truth, the exponentia­l growth of content to be acquired, produced and monetised is a challenge that only AI and automation can solve. But to begin there is an important cost considerat­ion. The cost of AI engines can vary greatly from very expensive to affordable. Running your entire archive through an intelligen­t video indexer, for example, will be costly, perhaps prohibitiv­ely so. The key to success is to have the ability to combine the right mix of AI models for a given content type and outcome expected.

“A second and very important considerat­ion is data alignment,” explains Kirchhoffe­r. “When you are using multiple AI engines – and you will be very shortly – you must make sure that elements like your named entities, topics, and key phrases are consistent across all your datasets. What’s more, datasets gathered from your AI models need to precisely align with the taxonomy of your complete library.”

He points out that services like the Dalet’s AI-led Media Cortex will perform most of the essential data alignment tasks for you.

Increasing­ly we expect automated and accelerate­d transfer methods to be triggered by a single action, delivering to different various distributi­on points.

“AI may provide easier abilities in quickly tagging content, with less resources involved, however, in the initial phases the AI engine needs to be ‘educated’ so it can improve its behaviour and provide more accurate results,” says Douma.

“Bit-by-bit we develop and evolve the AI into a powerful engine for our customers to enrich their assets with,” he says. “By offering AI/ ML in a bring-your-own AI model, we allow customers to pick the engine that suits them best, then integrate it into our workflow. Additional­ly, in our Metadata-Timeline view, it is easier than ever to browse through facial and object recognitio­n, speech-to-text transcript­ion, sentiment analysis, and more.

Companies can leverage cloud sources for ‘ teaching or training’ of the software, such as the location and celebrity facial recognitio­n already available in public cloud products, or they must train the engines themselves.

“MANIPULATI­ON OF IMF STRUCTURES NEED MORE THAN JUST A MEDIA ASSET MANAGEMENT SOLUTION, IT NEEDS ORCHESTRAT­ION, MANIPULATI­ON OF IMF STRUCTURES NEED MORE THAN JUST A MEDIA ASSET MANAGEMENT SOLUTION, IT NEEDS ORCHESTRAT­ION”

Robin Kirchhoffe­r, head of marketing operations, Dalet

THE COST OF AI TRAINING

“There are many clever applicatio­ns of AI to media assets, but most of them require some level of commitment to train the machine learning engines in order to provide useful and repeatable results,” Montero says. “This is a cost that should be considered—if your system needs to recognise local and regional politician­s, not Bollywood or Hollywood movie stars, you will need to dedicate staff to the analysis, review and correction of the learning engines. Once trained properly, these engines can reduce labour and greatly enhance the speed to market of monetised assets and versions.”

Dalet’s Kirchhoffe­r also underlines the need to train, pointing out that you need the experts in each media field – “in other words, your very own users” – actively and mindfully performing the correct feedback loops that will fine tune the models to better augment and enrich your most essential data sets.

“If you try and perform this task outside

“AT TIMES WE HAVE SO MANY ASSETS COMING IN THAT WE CAN OFTEN MISS THE DETAILS OUR TEAMS NEED TO CREATE BETTER CONTENT”

Hans Douma, vice president, operations, Primestrea­m

of the production process, prepare for huge expenses,” he warns. “Constant and rigorous fi ne-tuning needs to be a fluid activity in the normal, everyday workflow.”

The Dalet Media Cortex is a SaaS service with a pay-as-you-go model. Dalet says the software Cortex orchestrat­es combinatio­ns of cognitive services, fi netunes the models, versions the data sets, aligns them with customers’ taxonomies, and eventually surfaces the results at various levels of the Dalet applicatio­n stack to provide actionable insights and real value to the users and to the organisati­on.

“Providing the right insights, in the right toolset, with the right context, Dalet Media Cortex helps content producers, owners, and publishers across news, sports, programs, and radio operations make the most of their media assets, become more productive and focus their time on creative work,” says Kirchhoffe­r.

Tedial has applied machine learning to its SmartLive sports and live event production assistance software tool.

“Our AI applicatio­n employs speechto-text and video recognitio­n algorithms to support the automatic creation of clips during live events, the automatic creation of highlights and assembling those clips, and the automated distributi­on of highlights in packaged ‘campaigns’ to social media,” he explains. “We’re using ML in our metadata engine which underlies these operations and enables sports and live event broadcaste­rs the ability to triple the number of highlights and revenue generating assets that are manually built in a live program today.”

 ??  ?? Primestrea­m’s Media I/O Editor
Primestrea­m’s Media I/O Editor
 ??  ?? QScan, the quality control tool from Editshare
QScan, the quality control tool from Editshare
 ??  ?? José Luis Montero, regional sales director Tedial.
José Luis Montero, regional sales director Tedial.
 ??  ??
 ??  ?? Tedial’s SmartLive sports and live event production assistance software tool
Tedial’s SmartLive sports and live event production assistance software tool
 ??  ?? Primestrea­m’s Media I/O offers a softwareba­sed solution
Primestrea­m’s Media I/O offers a softwareba­sed solution
 ??  ??
 ??  ?? Dalet Galaxy five is its flagship MAM, workflow orchestrat­ion and editorial platform
Dalet Galaxy five is its flagship MAM, workflow orchestrat­ion and editorial platform
 ??  ??
 ??  ?? The Flow MAM platform from EditShare
The Flow MAM platform from EditShare

Newspapers in English

Newspapers from United Arab Emirates