By Shaun Lim
With more people staying at home because of the pandemic, video streaming continues to surge as the demand for content grows exponentially. This, in turn, is driving the rapid adoption of artificial intelligence (AI) and machine learning (ML) in the broadcast and media industry as a means to gain key insights into audiences, and to gain operational efficiencies through the automation of workflows.
One noticeable and growing use case for AI/ML, suggested Stan Moote, Chief Technology Officer of IABM (International Trade Association for Broadcast and Media Technology), is automatic speech recognition (ASR).
He told APB+, “Live captioning not only frees up people for other activities, it has the benefits of having a consistent delay of only a few seconds, whereas the delay for manual-captioning is variable. This allows for setting up captioning signal flows to be able to handle video switching at regional, national, local or network levels. Moreover, the accuracy is more than acceptable as using this for file-based activity is widely accepted.”
With ASR quickly becoming a commodity item, more and more use cases are emerging to differentiate product offerings, including using ASR collection of speech/script for translation.
Where script translation is concerned, Moote said having decades of news archives on tape is a perfect example of digitising video and creating a searchable script.
He explained, “News archives are used often, but only a tiny fraction of the actual storage gets repurposed.
“The issue with news archives is that it is totally unknown which older news clips are relevant until a current event requires older backup clips. Having the script simplifies the search process for the old news, which is often required instantly.”
Broadcasters and media companies are also finding out how AI and ML techniques are significantly improving the search process as often, each tiny regional area, or cities within the same country, can have unique key words and nuances. “Areas of improvement with ASR that are underway, include denoting a new speaker and creating speaker change IDs,” said Moote.
Remote work accelerating AI/ML deployment
As more people worked from home during the pandemic, indexing video using AI/ML also saw an increase because of a lack of direct and full access to content. “Finding mark-in and mark-out points through keyword searching, along with black and clip detection for top and tail, has become invaluable,” Moote observed.
He also believes that even as people gradually return to their work offices in 2022, the use of AI/ML for auto-trimming will increase as the time saved frees them up for other activities. “However, while
auto-trimming may be perfect for short-form content like ads, more work needs to happen for it to work for long-form content,” Moote pointed out.
He also observed that while AI/ML helped to repurpose old content to meet viewers’ insatiable demand during the pandemic, it is likely that in the future, AI/ML will be utilised instead, to re-cut older programmes into newer formats.
Content creators have also embraced the use of AI/ML, noted Moote. “I had the opportunity to interview a creative director during one of IABM’s BaM Live! Events about creating unique content for each type of viewing device, including issues such as how certain graphics do not work on tiny viewing devices in comparison to large screen TV sets. She told me that consideration is being done in the planning stage, after which AI takes care of the rest of the work – so basically hands-free!”
With ransomware attacks on the rise, AI/ML-based analytics also performs an important function in providing another layer of security monitoring, Moote highlighted. “Understanding when and why peak file and network activities are underway definitely helps with monitoring facilities.”
Key adoption drivers for AI/ML
As broadcasters and media operators begin to step up their operations and facilities for the future, they need to take into account how systems, automation and orchestration are becoming more and more complex. “Users will require assistance to operate them, as the average person cannot manage these on their own any more, in a practical sense,” said Moote.
Predicting that AI/ML will increasingly have a role to play in the media and entertainment industry, he added, “I believe AI and analytics have applications throughout the content supply chain, so the bottom line is whether it is about enabling more informed decision-making, or workflow automation and optimisation.
“This type of application and solutions liberate creative resources through routine workflow automation, solutions that automate and optimise content supply chain logistics and track its state, and solutions that optimise delivery and distribution, as well as enable more personalisation of the consumer experience.”
While the pandemic may have compelled broadcasters and media operators to invest in AI and analytics, a move that brought initial pressure on stretched budgets, Moote suggested that in the long-term, drivers such as demanding content pipelines will continue to drive increased spending on AI and analytics solutions in the media industry.
“Our industry now better understands that driving efficiency and personalisation through these technologies can differentiate themselves from the competition,” he concluded.
Question: What key role can AI/ML play in the media and entertainment industry, in particular the content supply chain in 2022?
Please send your views to firstname.lastname@example.org