Respeecher revoices
Darth Vader
The implications of AI in M&E
Machine learning: Who’s teaching who?
The award-winning Telos Infinity® VIP is the first fully-featured cloud-based intercom system. It brings sophisticated comms to any computer, tablet, or smartphone through its browser-based UI or VIP app for Android and iOS, making on-premises, cloud, and hybrid media production workflows easier.
Iwasn’t a great student while at school. I was doing ok and then sometime before my GCSE year I became obsessed with Steve Wright in the Afternoon on Radio 1. I stopped revising and spent my afternoons listening to the show. When I announced that I wanted to work in radio, my sixth form teacher told my parents I’d never get a job in the media. She was wrong.
Anyway, I got through my GCSEs and A-Levels, did an HND and then a degree (which I hated, far too theoretical). But as part of my dissertation (on the future of Radio 1) I got to know the Radio Academy and ultimately met someone who would have a huge influence on my working life.
continued to use that throughout my career. In the end, Howie would describe me as being like a terrier, I wouldn’t let anything go. It also developed my love of working on breaking news, which continues to this day.
I know how incredibly lucky I was. I have no idea where I would be 20-plus years later if I hadn’t met him.
The reason I’m telling you all this is because I firmly believe in mentoring and never telling someone they can’t or won’t succeed. That’s why a large part of this issue is dedicated to education and training in our industry. There are many organisations that want to help. Whether you’re just starting out or if you’ve been in the industry for a few years and are
www.tvbeurope.com
FOLLOW US
Twitter.com/TVBEUROPE / Facebook/TVBEUROPE1
CONTENT
Editor: Jenny Priestley jenny.priestley@futurenet.com
Graphic Designer: Marc Miller
Managing Design Director: Nicole Cobban nicole.cobban@futurenet.com
Contributors: David Davies, Neil Maycock, Robert Shepherd, Matt Stagg
Group Content Director, B2B: James McKeown james.mckeown@futurenet.com
MANAGEMENT
SVP Wealth, B2B and Events: Sarah Rees
UK CRO: Zack Sullivan
Commercial Director: Clare Dove
Managing Director, B2B Tech & Entertainment Brands: Carmel King
Head of Production US & UK: Mark Constance
Head of Design: Rodney Dive
ADVERTISING SALES
Advertising Director: Sadie Thomas sadie.thomas@futurenet.com +44 (0)7752 462 168, SUBSCRIBER CUSTOMER SERVICE
To subscribe, change your address, or check on your current account status, go to www.tvbeurope.com/subscribe
ARCHIVES
Digital editions of the magazine are available to view on ISSUU. com Recent back issues of the printed edition may be available please contact customerservice@futurenet.com for more information.
LICENSING/REPRINTS/PERMISSIONS
TVBE is available for licensing. Contact the Licensing team to discuss partnership opportunities. Head of Print Licensing Rachel Shaw licensing@futurenet.com
While at an event I met Howard Hughes, then newsreader on Chris Tarrant’s breakfast show on Capital FM. Two weeks later I started working as Howie’s news assistant and after a bumpy few months to begin with, I’ve never looked back. I’m so lucky that I found someone who has served as both my boss, my mentor and my friend. I left Capital after three years and moved to Ananova, an online news website, before returning to radio and working with Howie again, at LBC.
I always say that the time I spent working with Howie was my apprenticeship. He instilled in me a mantra, “think laterally Jen” and I’ve
reaching management level, there is help and guidance out there if you want it.
Our other focus in this issue is artificial intelligence, and it was interesting to hear about how AI is helping technology vendors answer questions from their customers during some of my discussions at IBC. There’s a lot to be wary of with AI, and David Davies covers that in the second part of his series looking at how it will impact the media industry. So here’s a question for you: if AI helps educate us, isn’t that a good thing? I'd love to hear you thoughts. n
JENNY PRIESTLEY, EDITOR @JENNYPRIESTLEY“I firmly believe in mentoring... I’m so lucky that I found someone who has been both my boss, my mentor and my friend”
David Davies explores some of the big ethical dilemmas surrounding AI, the prospects for meaningful regulation, and the implications for employment and the environment
We showcase the winning entries from the TVBEurope category of Future’s Best of Show awards programme at IBC 2023
Jenny Priestley meets #GalsNGear founder Amy DeLouise to hear more about the initiative and its aims to help women already established in the media tech industry become more visible
David Davies talks to the team at Respeecher, which, five years after its launch, is going from strength to strength despite the ongoing war afflicting its home country of Ukraine
Stephen Tallamy, chief technology officer, EditShare, discusses the importance of learning your craft
27
The virtual production skills gap led to a shortage of qualified and experienced professionals. Now, the UK’s ‘inaugural’ degree programme in the field aims to train tomorrow’s experts, writes Robert Shepherd
The wild ride of artificial intelligence and media and entertainment has begun, writes Kevin Riley, CTO and SVP of technology and innovation at Avid
Artificial intelligence and its potential impact on both the creative and technology sides of the media and entertainment industry was a dominant topic at the recent IBC show. Of course, this was no surprise. Companies across all sectors are wrestling to understand the opportunities and threats this technology brings.
In a previous article, I explored the potential of generative AI, and while it can be incredibly impressive, it does feel that the magic spark of human creativity is some way off. The term ‘derivative’ is possibly too strong, but AI-generated content is fundamentally derived from what has gone before and that exposes a potential issue; that of intellectual property.
When we use a tool like ChatGPT the best results are achieved through an interactive process, asking questions and then iteratively modifying the results through further questions or changing the emphasis of the result; e.g. rewriting the answer using less formal language.
As we go through this our focus may be on getting the results we want, but of course, we are dealing with a system that learns from its interactions, so we are also training the system on how to produce an optimal response. It may feel like we’re learning from the AI interaction, but I think it’s important to recognise that we are simultaneously both student and teacher.
This is almost certainly why tools like ChatGPT are free to use, not due to a philanthropic cause for the greater good, but because training AIs is a lengthy and potentially costly process. Also training an AI model requires imparting knowledge and intellectual property to the AI system. OpenAI, the company behind ChatGPT, was created with the stated intention to develop “safe and beneficial” artificial general intelligence, and it was founded as a non-profit organisation. This all sounds great, but subsequently, OpenAI has transitioned to a for-profit organisation, albeit with some restrictions. This was primarily to raise more investment, including venture capital, but the company has also announced its intention to commercially licence its technologies. Personally, that makes me feel more cautious about what I might share with an AI system.
Amazon and Apple are among the major companies that have placed restrictions on the use of tools like ChatGPT. Concerns
range from the security of customer data to the loss of proprietary company information. Using AIs to help write software code is a great example; once you’ve worked with an AI to solve a coding problem it can of course solve that same problem for anyone else. We also need to consider what a completely open community of users might be teaching an AI, deliberately or inadvertently. An AI system, like any other computer, will suffer from the old adage ‘garbage in, garbage out’, or a more modern and nuanced dynamic is the social media ‘echo chamber’ where users can get validation of their views because they follow like-minded people. The information we receive is based on criteria and information provided to the algorithms. The complexity of AI can take this to another level, where we don’t necessarily understand what we are teaching the AI model.
An ex-colleague and friend of mine James Cain has a perfect example of this. For his PhD, he was developing an AI model to recognise World War II tanks. The model was trained with a series of pictures of friendly and enemy tanks. However, when James came to test the model it failed spectacularly. He eventually determined that the pictures of friendly tanks had all been taken on grey days, and the enemy tank pictures had been in sunny weather. So, the AI model had actually learned to read the weather and not the tanks. With the advanced AI models of today, there are some high-profile examples of unintended bias that have been created; a real challenge for the technology going forward.
As a conclusion to this article, I had intended to conduct a small experiment and have ChatGPT write one paragraph and then ask you to determine which one it is. However, when I tried I think experienced a form of bias. The tone of my piece has been somewhat cautionary about AI, and when I asked ChatGPT to write a paragraph on companies placing restrictions on the use of AI the results were interesting.
Phrases such as “several forward-thinking companies have implemented restrictions on employees”, and “several prominent companies have recognised the importance of ethical AI usage”, were overtly complementary to the companies involved. I think it would have been extremely obvious which paragraph I hadn’t written myself, but it’s interesting to reflect that when using AI we need to be careful to deliver our own message and not someone else’s! n
Lord Baker informed me in Summer 2022 that the eight secondary school subjects announced by the government 100 years ago are the same eight subjects still being taught today. I often reflect that our education system is not fit for purpose, and this reinforced my thinking.
Schools and teachers are under such continuous and enormous pressure – the current RAAC [reinforced autoclaved aerated concrete] situation just being a small example of the various factors that must be juggled on a daily basis – that it is no wonder they often struggle to think beyond focusing on curriculum delivery. Managing exams, grades, behaviour, and truancy (all exacerbated by Covid) means that embedding careers advice and providing exposure to the ever-growing choice of job roles is often far down the list.
Simultaneously, the broadcasting industry is facing a skills crisis across multiple roles, and we continue to suffer from a lack of diversity; yet we still question why things are not aligned and the issues remain.
I have seen first-hand over the last four years how teachers often work in run-down conditions, where children aim to just get through school. Many are not remotely thinking about a career in the media industry or any form of post-18 education pathway. Even if they are (and evidence suggests that many young people would consider a job in the industry), the route to getting into the sector seems over the rainbow and very far away. The social and economic barriers and systemic environments that exist are too strong and ingrained for them to even consider that a media career is possible. I believe we have a duty and responsibility as an industry to reach these young people and to inspire them about this dynamic sector. We cannot rely on individual teachers or schools to solve our issues; there is brilliant and diverse talent out there, we just need to go out and find it. And, critically, invest and support it.
We need to be much clearer about the pathways into the industry. If you were advising an 18-year-old today, what would you tell them to study? Where would you tell them to go? A
degree, a diploma, a BTEC (which are due to be cut next year)?
The forthcoming ‘T’Levels (if the government doesn’t cancel or postpone these again)? What about if they wanted to enter the industry at 18; can they? Which companies accept 18-yearolds? There is such a lack of clarity it’s no surprise that teachers, parents, and young people struggle to navigate the path to a career in media.
That’s why I set up the Media Careers Podcast; to inspire and inform educators and students about the breadth of roles across the industry. With 55 per cent of 12-34-year-olds listening to podcasts on a monthly basis, the podcast focuses on industry professionals’ education and routes into the industry and hopefully allows young people to hear themselves reflected in some of these stories. Whether that’s Sean Williams discussing his dyslexia and how it has enabled his career development or Hope Primus discussing her 24/7 commitment to getting a role in the industry.
In addition to this, included in the show notes are direct LinkedIn connections to the speakers, which means teachers can reach out to ask the guests to visit their school, and young people and those interested in working in the industry can find out about internship opportunities. There are also details of other organisations and educational institutions that can provide further support. Hopefully, it can help to break down barriers and the long-held belief that ‘it’s not what you know, but who you know’.
Finally, there are also lesson suggestions for teachers to use in the classroom as a follow-up to the podcast, building on the information shared within each episode; for example, ‘practice timing a script or writing a recording a three-minute video of yourself showcasing your skills and talents that an employer might be interested in.’
Collectively, we still have so much more to do to ensure we have a diverse workforce and to see the skills crisis addressed; the Media Careers Podcast hopes to help in a small way, and we welcome hearing from industry organisations interested in getting involved or supporting the podcast as we expand our work into live events and internationally. n
As we journey through the ever-evolving landscapes of 5G, multiaccess edge computing (MEC), and the dynamic realm of artificial intelligence, a captivating synergy comes into sharp focus. This fusion holds the promise of propelling the broadcasting industry into an era where the amalgamation of these technologies transcends individual capabilities. In the fast-paced realm of contemporary media production, where unwavering efficiency and uncompromising quality reign supreme, a triumphant trifecta of technological forces stands poised for a revolution: 5G technology, the realm of private networks, and the exhilarating frontier of edge computing. These forces teeter on the precipice of transforming every facet of media production.
In the realm of live broadcasting, 5G’s transformative power shines brightly. Its ultra-fast wireless connectivity empowers untethered cameras to transmit high-resolution footage swiftly, crucial for dynamic live productions. Especially in remote or mobile settings where fibre installation is impractical, 5G and private networks offer unmatched reliability and security. These dedicated networks ensure seamless realtime communication between production teams, irrespective of location, while AI enhances content quality and delivery. The convergence of 5G, private networks, and AI revolutionises live broadcasting, making it more mobile, efficient, and accessible, propelling the industry into an era of unparalleled possibilities.
The substantial bandwidth capabilities of 5G are a game-changer for live video streaming, and when coupled with edge computing, the possibilities expand into thrilling territory. Edge computing entails processing data closer to the source, eliminating the need for data to traverse long distances to distant data centres. In media production, this translates to complex video processing tasks, such as real-time video encoding, being executed at the network’s edge. AI plays a pivotal role in this transformation, ensuring that video content is not only streamed seamlessly but also optimised in real time.
An era of remote production like never before is being ushered in by 5G private networks. Infused with the reliability and security of private networks, production teams gain centralised control over cameras, lighting, and audio equipment, all while AI algorithms could monitor and optimise each aspect of production. The true magic of this convergence lies in the integration of edge computing, where real-time video editing is executed with AI precision, enhancing the quality of live broadcasts or recordings without the need for extensive post production work.
The low latency and ample bandwidth inherent in 5G, when harmonised with edge computing, create an ideal playground for augmented reality (AR) and virtual reality (VR) applications in media production. AI algorithms power these immersive experiences, enabling creators to craft virtual sets and interactive storytelling. Edge computing ensures that the processing demands of AR and VR are met locally, guaranteeing a seamless and responsive experience for both creators and audiences.
Edge computing’s contribution to enhanced security in media production is complemented by AI’s prowess. By processing data locally at the network’s edge, sensitive content remains protected within the secure confines of the private network. AI-driven content analysis operates in real time, swiftly detecting and filtering out inappropriate material, ensuring a secure viewing environment.
The internet of things (IoT) is a vital component of modern media production, with numerous connected devices used on sets. Edge computing, coupled with AI, facilitates efficient monitoring and control of IoT devices, ensuring seamless operation during production. It also allows for the rapid analysis of IoT-generated data, providing valuable insights that optimise production processes.
While the adoption of 5G, private networks, edge computing, and AI initially demands an investment, the long-term benefits in terms of cost efficiency are substantial. Reduced travel expenses, heightened productivity, accelerated turnaround times, and optimised resource utilisation all contribute to lower production costs and heightened profitability.
The integration of these technologies into the world of media production marks a significant leap forward in the industry’s capabilities. From ultra-fast connectivity and low latency to high-quality streaming, remote production, and the innovative realms of AR and VR, AI emerges as the driving force behind this transformation. And as the technologies continue to evolve and synergise, the future of media production gleams with promise and excitement.
Content creators, studios, and broadcasters that wholeheartedly embrace these advancements not only streamline their workflows but also deliver richer, more captivating content to audiences worldwide. In doing so, they set new benchmarks for the industry.
These transformative technologies, with AI at their core, usher in an era where creativity knows no bounds, and efficiency reigns supreme. n
TVBEurope’s website includes exclusive news, features and information about our industry. Here are a few of our recently featured articles…
Rugby World Cup 2023 is employing a number of broadcast innovations, including uncompressed end-to-end IP stream contributions between venues and the IBC; new cine-style cameras; and more access to the teams than ever before.
TVBEurope spoke to Host Broadcast Services to find out more.n
A poll by TVBEurope has found that most people within the media technology industry believe innovation and technology have the biggest impact on the sustainability values within their organisation. n
TVBEurope speaks to a cinematic lens manufacturer, service provider and leading sports broadcaster to get their thoughts on why cinematic technology is seeing broader adoption, how it enhances the viewing experience, and how the trend will continue. n
We talk to editor Alex Fountain about his work on the BBC’s new drama Boiling Point, and how he’s followed the project from short to feature to TV. n
Editor Michael Brown explains how he goes about constructing each 30-minute episode of the hit documentary series, and the technology that helps him do it. n
Can we close the stable door, or has the horse already bolted? Headlines to this effect have been a recurring feature of newspapers and current affairs programming throughout 2023 as the phenomenal – and, it seems, potentially catastrophic – implications of AI have become more apparent to a wider public; or in other words, all of the people who are not in the crucible of this technology’s development or deployment.
In the first part of TVBEurope’s exploration of AI, we focused on some of the fastest-growing media and entertainment use cases, including creative assistance and sports commentary. The benefits to these kinds of applications – which, to date, tend to be centred around providing content where it isn’t currently provided by humans – are hard to contest, but there is no guarantee that it will stop there. So it’s no wonder that sections of M&E – as epitomised by the screenwriters and actors’ strikes, in which AI has been a factor – are now locked in increasingly fractious debate.
This summer’s protracted strikes have given voice to many creatives’ most urgent concerns about the present moment. For writers, the anxiety revolves around potential loss of authorship to, or even outright replacement by, AI; for actors, it’s primarily about losing control of their likenesses or replication without renumeration.
Simultaneously, there has been growing clamour for meaningful regulation, and it is here where the fears that it might already be too late are most acute. Nonetheless, there has been a lot of activity in this area over the last few months, including the EU’s approval in June of the draft text for what it describes as “the world’s first comprehensive AI law”. Currently out for discussion with EU member countries with a view to reaching agreement by the end of 2023, the Act will establish obligations based on the level of risk and ban outright those applications – including real-time facial recognition systems – that it deems to be ‘unacceptable’.
“[Companies should] be upfront with employees with what they envision and think about AI. Transparency will reduce fear and speculation among the workforce”
JOHN FOOTEN
Of most immediate interest to creatives will be the proposals surrounding generative AI, where transparency requirements would include “disclosing that the content was generated by AI, designing the model to prevent it from generating illegal content, [and] publishing summaries of copyrighted data used for training”; a measure that would seemingly address the concerns of the 17 authors, including John Grisham and Jodi Picoult, who are currently suing ChatGPT developer OpenAI.
While these moves towards a formal legal framework are generally welcomed, there is also a sense that they need to be complemented by rigorous technical, M&Eoriented standardisation. “As with any other kind of technical development, regulation and international standardisation are very useful, and are actually a must for the healthy development of the industry,” says Michael Yang, senior vice president, corporate development, Caton Technology.
Chris Bailey, head of innovation at broadcast SI and service provider Jigsaw24 Media, suggests that the issue of deepfakes goes to the heart of regulatory challenges within M&E. “There’s no question that we need to be able to protect people from having their likeness used without their consent or fair compensation, but we also don’t want to stifle the development of the technology behind this capability,” he says. “The challenge is to find a good ground between creating legislation that protects people while still allowing AI capabilities to flourish and develop.”
Long-term industry observer and executive John Footen, managing director, media and entertainment at Deloitte Consulting, predicts that standardisation will emerge from the adoption of more focused AI modelling: “In my opinion, the future will likely be populated with smaller and more targeted models that perform narrower functions more consistently and effectively. Real-world workflows will be populated by multiple AI entities that will need to communicate effectively in a way that creates reliable results. Like other software systems, this will come about more easily with some form of standardisation.”
But while legal and technical frameworks should limit some of the more potentially damaging outcomes for M&E, they are unlikely to do much about addressing another burgeoning fear: that some roles might be replaced altogether, and relatively soon, leading to largescale redundancy of technicians. But although some degree of unemployment appears certain to occur, new opportunities will also be created.
“The biggest fear about AI is whether it’s going to take people’s jobs, and I think it absolutely will,” says Bailey. “If you take VFX rotoscoping as an example, I think AI will make a massive dent in the people power that’s required to do this kind of task. However, the people doing this job were always low paid, worked long hours, didn’t necessarily get the recognition they deserved, and could potentially be more useful concentrating on something else.
“The flip side to the fear that AI is going to take over some jobs is that there isn’t a replacement for the human at the end of the chain. AI is only as good as the information you feed into it, and there will still need to be some sort of human involvement in every creative process.”
For Footen, there is a comparison to be drawn with the transition from physical to file-based media. “There were many predictions about job losses and other implications at the time,” he points out. “What wasn’t discussed enough was that the jobs would change, not go away, and that the media product that we created would get more complex. Think of the relative complexity of a sports broadcast experience in the 1970s versus today. It is now far more complex and has entirely new jobs with new skills that didn’t exist decades ago.”
He suggests that the situation with AI is similar, but urges media companies to “be upfront with their employees regarding what they envision and think about AI. Transparency will reduce fear and speculation in the workforce, [whilst] a plan to adapt your talent to the changing environment will give people clarity on how to evolve their careers.”
One might hope for similar transparency regarding the environmental impact of AI, whose immense processing requirements will add to a demand for data centres already being greatly swelled by cloud adoption.
“The more people use AI,” notes Bailey, “the more CPU and GPU power is required, the more data centres are needed, and the more power is being consumed to run and cool these servers. I don’t know how much processing power OpenAI is planning to allocate to ChatGPT in the future but, with current use already reported at up to 30,000 NVIDIA GPUs, I bet it will be biblical.”
In the wider world, the scale of the challenges around AI confronting governments worldwide – encompassing everything from potential takedowns of major infrastructural systems to the criminal deployment of bioweapons – are monumental and arguably without historical precedent. But even within the parameters of our industry, we ignore them at our peril, and only by confronting them head-on now do we stand a chance of a long-term, harmonious co-existence with AI. n
Future’s Best of Show awards programme at IBC 2023 once again highlighted the amazing innovation going on behind the scenes in R&D and product development facilities across the industry. Here, we showcase the winning entries from the TVBEurope category
Judges comment: “Sustainability is a major watch word in broadcasting today, so any platform that makes it easier for companies to find eco-aware products and systems should be welcomed.”
Judges comment: “I was aware that Agile Live had already been deployed successfully for some high profile events, so evidently it is a proven and capable solution. This product’s ability to move the time domain is unique to my knowledge, [and] enables production methodologies not possible otherwise.”
Judges comment: “Sold as an easy to use platform with a rich feature set to encourage creativity, this seems like it promotes engagement in a variety of ways, while taking care of timing and connectivity in the background. It looks pretty cool to me.”
Judges comment:
“Easy to use UI for a solution that appears true to its word in helping broadcasters streamline workflows and automate some of the more repetitive tasks therein.
BluSpot frame-accuracy tool significantly reduces manual (frame-byframe) ad marking analysis.”
Wireless IP and Low Earth Orbit
Satellite TV Production Demonstration
Judges comment: “A very exciting wireless IP demo from BT. Throughput on Low Earth Orbit isn’t perfectly constant, so the use of two different constellations provides an innovative solution.”
Judges comment: “This seems to be a very versatile cloudbased editor that will appeal to a wide range of professionals, working across a broad spectrum of content. A genuinely good product which seeks to differentiate itself amongst some similar competitors.”
Judges comment: “As the environmental and cost implications of largescale data centre-based cloud deployments become clearer (and I think that is happening in 2023), it’s likely we will see more alternative solutions. So, I think the NEO series could do very nicely for Appear.”
Judges comment: “Hard to see what this entry doesn’t do! IP67 rated and rugged, it looks like it would cope with whatever a client needed in a variety of harsh environments, with multiple connectivity options.”
Judges comment: “12G-SDI is in extensive use across broadcast, especially for OB/live applications, and that will continue to be the case for a long time; especially in environments where a complete shift to IP is not currently viable. I can see this new 12G-SDI router being very popular.”
Judges comment: “A slew of major new features for the Dina newsroom system: impressive.”
Judges comment: “IP-based SaaS services are in the ascendant, so backed by the name of Evertz I would expect this to achieve good traction pretty quickly.”
Judges comment: “There is a clear demand for video encoder/ transmitters capable of working with the new 5G network infrastructures as well as the existing 3G and 4G ones, so I would expect this solution to do very well.”
Judges comment: “The latest version of Harmonic’s acclaimed Media Software shows that the media technology innovator is continuing to keep pace with the many production & delivery changes occurring across the industry.”
Judges comment: “With digital streaming exploding, more devices, cheaper data and better connectivity, this technology looks long overdue. If it can reduce the carbon footprint of companies in the broadcast media and streaming sector by up to 30 per cent with no reduction in service, why wouldn’t they implement it? Not seen anything like this before and looks easy to set up and implement. My favourite entry this year!”
Judges comment: “A very flexible monitoring system that seems to fill a rather notable gap in the marketplace. With the Genelec name behind it, I am sure it will be received warmly by users in broadcast, post, music and gaming audio.”
Judges comment: “This looks to be a very sophisticated approach to the increasing challenges of high-volume, large-scale data orchestration.”
Judges comment: “Virtual production has taken off rapidly, but for some companies has entailed too much cost at launch to fully engage with. Affordable VP solutions like this are therefore bound to resonate with that group of users, in particular. It’s a brilliantly packaged system which reduces the difficulty and cost of a growing market segment.”
Judges comment: “A strong entry because it sets a data limit based on the customer experience and reduces the amount of data needed to deliver content to maintain the same quality. It is invisible to the consumer, but it reduces costs to the CDN as much as possible. It also appears to have flexible deployment and is ‘set and forget’, so it runs in the background with no changes required to existing workflows.”
Judges comment: “Captioning and QC are undoubtedly two of the biggest growth areas for AI/ML applications right now, so this seems like a very clever and well-timed introduction.”
Judges comment: “LiveU has a very distinguished track record of developing live production and contribution solutions, and this new product would appear to be entirely in keeping with that. In particular, the emphasis on ‘overcoming complexity’ associated with IP workflows sounds like a winner to me.”
Judges comment: “A multi-faceted analytics tool that will help media marketeers optimise their marketing strategies. Sounds like a winning solution in a sector that will only become more competitive.”
Judges comment: “An asynchronous architecture seems like a big step towards freeing up cloud production, and that it is already cloudnative also reads like a statement of intent. It does seem genuinely disruptive and flexible.”
Judges comment: “A tool that provides more comprehensive identification and background information about non-scripted content seems overdue, so this innovative offering from Media Distillery is to be welcomed.”
Judges comment: “Delivering content localisation, and at huge scale, is a very pressing challenge for a lot of media companies. LTN Wave feels like a very good answer to this challenge.”
Judges comment: “This appears to be a very innovative solution that will allow a broader spectrum of content owners to optimise monetisation; a critical consideration in an intensely competitive sector.”
Judges comment: “Broadcasters and media companies are at varying stages of their journey to the cloud, and that is likely to be the case for some time. So a product that can accommodate all of these modes of operation is bound to resonate.”
Judges comment: “A lot of scope for a (re)configuration of solutions like this, especially with so many fast-turnaround productions. TFC is a bold internal product which has become valuable enough to take to market. It is a difficult thing to create and requires a lot of understanding to do it well.”
Judges comment: “This appears to be a good, versatile solution for a range of AV over IP installations, including those that involve thousands of endpoints.”
Judges comment: “This solutions is well-explained and should resonate strongly with the ongoing shift towards ST 2110/AES67 workflows.”
Judges comment: “Ross Video has been diversifying into a number of areas, both in broadcast/media and pro AV, and this looks to be another bold and innovative move by the company. Making VP more efficient and less complex is a driving force in the industry, too, so that’s another reason it should do well.”
Judges comment: “Another potentially useful tool for the ever more complicated task of finding specific material amongst a huge volume of content.”
Judges comment: “A versatile rasterizer that can work with both legacy and new/emerging workflows is bound to be in-demand, and in many areas. PHABRIX is the market-leader for such solutions; hence, I am sure it will do very well.”
Judges comment: “The growth of streaming and demand for channels is calling for the ever faster implementation of new platforms and services. This system certainly looks to offer features for rapid creation and control of channels.”
Judges comment: “An innovative approach to providing greater capacity and flexibility for a demanding streaming market.”
Judges comment: “Cloud operations are now firmly part of the broadcast and facilities sector and Telestream appears to be offering a wide range of functionality to allow users to not only get the most out of the ‘new’ platform but also bridge the gap to their premises.”
Judges comment:
“Efficient asset management is essential in modern broadcasting. The influence of AI is set to push this area of technology into new areas, making it easier and quicker to find and identify content.”
Judges comment: “Compact and versatile, this should do very well for Vislink.”
Judges comment: “Audio has always been a key component of broadcasting and, in today’s world of podcasting and audio books, that status is now more widely recognised. The MPA1-MIX-NET-V-R will be important in ensuring that broadcast sound is not only of the best possible quality but that it is going where it should.”
Judges comment: “A potentially significant step forward in content search features. Selecting a programme according to mood or interest could either be a handy time-saver or something of a novelty. It is to be hoped it will be the former. Impressive product; its ability to understand video and audio in multiple ways allows it to create a lot of value for consumers and the broadcasters/ streaming companies that integrate the technology.”
Judges comment: “An impressively versatile streaming platform-as-aservice offering that is destined to do well with both new/ emerging and wellestablished service providers.”
Judges comment:
“The D2C Video Gateway is designed to give operators the flexibility to rapidly onboard video channels delivered in any protocol, over any network, process into any format, and deliver to any target. An intuitive and effective solution for broadcasters.” n
Launched at NAB Show in 2016, #GalsNGear is a movement that aims to bring equity to women in media and entertainment. While organisations such as Rise work with students to introduce them to the industry, #GalsNGear focuses on women already established in their working lives, whether they’re in the industry or thinking of joining.
The idea for #GalsNGear came to founder Amy DeLouise while in the ladies’ room at NAB. “I was a speaker and I had been on the show floor all day, and I thought, ‘how can I possibly be alone in this bathroom?’ There was no line at the ladies’ room and it kind of ticked me off,” she explains.
“Like a lot of women at many industry trade shows, and I’m not
slamming one particular show, we find ourselves as the one woman on a panel or one woman presenting on a technical subject. So, I said to myself, ‘I know lots of women who can do that, why aren’t they here?’”
#GalsNGear started with the need for a sense of belonging, says DeLouise. “Women would tell me, ‘I don’t know if I want to go to these big shows, it just doesn’t feel like a place where I belong.’”
This led her to approach the leadership of NAB Show, and the movement made its debut at NAB 2016 with networking events and a live stream from Central Hall lobby, with panellists discussing technical subjects and gear including cameras, lighting, and sound given away to women at the show. “The hashtag went viral, so it was clear that there
was a need for some kind of collaborative environment for women to talk about technology, meet each other, and collaborate with each other.”
In the intervening years, #GalsNGear has grown to include a number of different programmes. This summer it launched the Tequity Hub which includes virtual meet-ups on ‘Tequity Tuesdays’, designed to include discussions and demonstrations around new technology. “We also have a leadership programme because I really feel that women are missing from the management pipeline in terms of staying,” adds DeLouise. “Companies are getting women to join, but they’re not retaining them. Why is that? And they’re definitely not retaining women of colour.”
One key area that DeLouise hopes #GalsNGear can tackle is what she describes as “soft skills” that help women stay in management. The idea is to focus on four areas of leadership training: leading teams and leadership style; developing a leadership brand both within a company and the wider community; negotiation skills; and financial literacy. To empower women in the four key areas, #GalsNGear hosts live in-person training sessions several times a year, as well as virtual events on its Tequity Hub.
For those looking to enter the industry, there’s a student programme. This year, #GalsNGear is partnering with Chyron to teach the company’s Chyron Live software, which will include live coverage at NAB 2024 by students from the Walter Cronkite School of Journalism mentored by a #GalsNGear team. “We also have our beta testing network, and we function as an informal speaker’s bureau because people come to us to make sure they have women at high levels speaking at panels,” adds DeLouise. “Networking events are a part of this, as well. We try to make sure that we’re not necessarily doing what other groups are doing. We’re trying to fill in the gaps where we see things that might be missing.”
While we’ve been discussing the group’s US activities, it is active outside of North America. At IBC in September, #GalsNGear hosted a number of events, including joining forces with other advocacy organisations to organise a flash mob to highlight strength in numbers at the show. “We are a little different than a lot of other organisations,” states DeLouise. “We call ourselves a movement and not a membership organisation. There’s no membership fee. We let women self-select to be part of our community, and so we welcome people from all over the globe.
“It definitely started in the US because I’m in the US, but we just said, ‘join our movement, be part of this and find other women’. Wherever it is you’re going, whether you want to have women on your crew, whether you want to showcase your skills and be on one of our panels, or whether you want to advance your own career by taking one of our leadership deep-dive training programmes.”
DeLouise is keen to stress that visibility of women within the industry is key, and she sees it as being part of the equation behind the new Tequity Hub. “It’s visibility: access both to the technology and to the networks that make it possible to build your career; and then a community surrounding you to make you feel like you can achieve what you want to achieve. So it’s really a combination package that we think is important for women.”
#GalsNGear began its virtual programming during Covid when it was forced to move a leadership event online. The organisers realised that the audience was staying on the website for many hours, which led them to developing more “sticky” content. “We have monthly meet-ups with a Q&A on a particular topic. We’ve talked about how to become a beta tester
and why that’s important, as well as opportunities for women in sports broadcasting,” explains DeLouise.
The hub also lets the audience break into other areas, including a new skills zone. “I mentioned Chyron and we have a Chyron Academy with all kinds of amazing tools that you can actually use to upskill yourself in various areas. Riedel has similar training in the audio engineering area and Blackmagic has some for cameras, colour grading and editing.
“A lot of these companies have training tools, but we’re trying to pull them into one place. Sort of a one-stop shop so that women who are already in the industry can upskill, or if they want to shift careers,” she continues, “there are also job links in there. We’re working with a lot of different industry partners who want to be sure their jobs are visible. We’re very focused on women who are in the industry; they can be very senior but feel like they’re facing some kind of barrier to reaching the next level. We want to help them get there.”
All of this content is free. There are no membership fees for joining #GalsNGear, and DeLouise says she hopes they will never have to charge. “We’re trying to get more sponsors to join us. We have quite a few as it is but we always want more because we just feel like the industry deserves to be an equitable industry,” she continues. “There are dozens of jobs that didn’t exist when I started. So I really don’t see equity as a zero-sum game because there are just so many jobs to be covered and we need more people. I think the more training we can get, the more people of colour, the more women or people of every gender and background, the better it is for the industry to move forward.” n
Five years after its launch, AI voice cloning software company Respeecher is going from strength to strength, despite the ongoing war afflicting its home country of Ukraine, as David Davies reports
From a simple hunch or some playful experimentation, innovative and successful businesses are so often sprung, and Ukrainian AI voice cloning software developer Respeecher is no exception. Rewind to 2015 and two of the future co-founders of the company, Dmytro Bielievtsov (also CTO) and Alex Serdiuk (also CEO) were participating in a hackathon and experimenting with a novel idea about speech synthesis.
“We thought it could be fun to play around with voice and machine learning to see if we could make it sound like a different person,” recalls Bielievtsov. “Although we quickly realised that it was a lot harder than we had thought!”
Nonetheless, the idea remained with them and – a few years later and with some surrounding technology gaps having been addressed –Bielievtsov and Serdiuk were joined by fellow co-founder and head of
research, Grant Reaber, to establish Respeecher. “It had become clear that an avalanche was starting to form,” recalls Bielievtsov of the growing momentum around AI innovation at that time. “We saw it was the right time to form a company to explore the use of AI in voice audio.”
The result is a core technology that is able to replicate voices for media projects including films, television series and video games. Combining ‘classical’ digital signal processing algorithms with proprietary deep generative modelling techniques, the Respeecher workflow is structured around five primary phases: securing permission from the target voice; collection of a high-quality recording of the target voice (which can be pre-existing or newly made); collection of a high-quality recording of the source voice, although this may not always be needed; the use of AI to create pitch-perfect voice-to-voice swapping models; and finally, to get the desired voice content, speaking
into a microphone and then have the conversion take place using the aforementioned processes.
As well as its bespoke AI technology, Bielievtsov indicates that the company has taken a different tact from others who have explored similar terrain. “There have been some businesses who opted to concentrate on optimising scalability and resource efficiency, and wanted to go B2C right away,” he says. “But we decided that was not the main thing for us. Instead, we wanted to focus on audio quality and making sure that the rendered result was indistinguishable from real speech.”
A glance through Respeecher’s credits strongly suggests that they have achieved their goal. Most prominently, the company worked with LucasFilm to recreate the voice of Darth Vader, as provided by legendary actor James Earl Jones, from Star Wars: A New Hope. As reported in TVBEurope last September, Jones agreed to his voice being recreated with AI in the TV series Obi-Wan Kenobi. (The company also recreated Mark Hamill’s voice for another Star Wars TV series, The Book of Boba Fett.)
Acknowledging that Earl Jones’ voice inevitably changed as he got older, the decision to use AI reflected a desire to preserve the defining characteristics of Vader’s voice. While Bielievtsov is unable to discuss the company’s work on Star Wars projects in detail, he does pay tribute to the patience of the Skywalker Sound team as the underlying technology
was finessed. “I am super grateful to them for how tolerant they were of the imperfections in our technology, which was way less refined when we first started working with them in 2019 than it is now,” he says. “It really took a lot of faith on the client side for them to work with us at that point and not get frustrated by some poor initial results, but instead push on through to where [it was working very effectively]. We had not expected that people at a big studio would be so cooperative and willing to use a technology when it was quite raw and glitchy, so it was really one of the greatest experiences we could have had.”
Providing a sharp contrast to this rewarding process was the background against which the company’s later work on Obi-Wan Kenobi was conducted: the invasion of Ukraine by Russia that began on 24 February 2022. “We were delivering parts of the project from bomb shelters,” he recalls, adding that the ongoing conflict has been “very hard psychologically, although thanks to the army we are in relative safety at our centre in Kyiv. The war is still there with cruise missiles attacking cities, air-raid sirens and drones, but objectively there is not much physical threat now to our team.”
While the tentpole film and TV productions remain its core activity, Bielievtsov says that Respeecher has also taken moves to “democratise the technology” through a B2C platform, Voice Marketplace, that
doesn’t involve a direct service involvement. “This allows you to make and submit a recording that can be rendered according to one of the voices in our library, which now includes more than a hundred,” he explains.
The service has proven popular with sound professionals “needing to generate extra audio to be added during post production,” as well as game producers for various voiceover requirements, says Bielievtsov. “There has also been a lot of interest from business clients who want to localise advertisement voiceovers so they suit different markets,” he adds.
Aware of the need to support content authenticity in an increasingly dynamic area of technology development, Respeecher is also collaborating with several high-profile industry initiatives, including
the Adobe-led Content Authenticity Initiative and the Committee on Synthetic Media, where it’s partnered with DEG to work on a code of conduct for companies using AI-generated content and voice cloning technology.
For Respeecher, the focus for the next few years at least is likely to be on the continued development and dissemination of its core AI audio platform: “It’s our technology that we have developed over time and we are very proud of it. Plus our R&D team keeps developing new applications for it,” states Bielievtsov. But that doesn’t mean that some diversification might not take place further down the line: “We are partnering with other companies who are doing a great job with visuals, but at some point it might be good to do those ourselves and be able to offer everything in-house as a complete package.” n
“I am super grateful to [LucasFilm] for how tolerant they were of the imperfections in our technology, which was way less refined when we first started working with them in 2019 than it is now” DMYTRO BIELIEVTSOV
The difficulties of making our industry attractive to young people, then giving them the required education and development programme, have all been discussed regularly. But I see another level of complications ahead of us, and I believe we need to start the debate now.
Some people start as runners; some start with a degree at the growing number of universities that are providing media education, many of them excellent. But they all meet at the same point: they become assistants in their chosen specialisation.
These are real, important roles. Editors need edit assistants to take on the preparation work, organising content into bins and so on. The structure applies across the industry: script supervisors have assistants; composers have assistants; colourists have assistants. These roles are important to get the work done in a timely manner. And – this is very important – they give the assistants a real understanding of what the job is about.
What the assistants gain is a real, deep understanding of what good work is, and how it is constructed. What are the steps involved in making great work? How do you navigate the pressures of time and budget and other steps in the production workflow to achieve what you need to do to be the best editor or script supervisor or whatever.
Those who have gone down the degree route, and will often have had the opportunity to do their own projects with professional-grade equipment, will not have had the stimulation of working with someone who has been doing this job for real for many years. They will not have been subject to the time and budget pressures which are critical in the real world.
For anyone who went to IBC this year, there was one mantra heard everywhere: ‘AI is set to transform the industry’. Setting aside for the moment just how realistic it is to depend upon AI – which is not and can never be creative, just analytical of existing work – this idea has a potentially catastrophic effect on this unofficial apprenticeship system. A benefit of AI is that it can automate the dull, repetitive tasks traditionally offloaded to assistants. But if these tasks are taken away from assistants, then how do people make the transition from carrying cappuccinos (or finishing the degree project) to fully-fledged members of the profession?
On a major drama series production, the director and script supervisor realise that they are in danger of running over time, so they need a scene to be tightened, without losing key material but taking it up a pace. Today, that task could be handed to the script assistant, who gets terrific experience in creating dramatic dialogue, learning about pacing and – through working with actors and director – understanding what works and what is not practical or convincing to say.
Or the script supervisor could ask ChatGPT to do it.
When you are scoring a movie or a premium television series, the headline composer – like Hans Zimmer or Debbie Wiseman – will create a suite of themes, musical ideas which convey each character, along with sketches on how they will be used to convey different aspects of the drama. This is how music adds to the audience’s understanding of what is happening.
If you need a few seconds to underscore a particular scene, or you need to adapt an orchestral idea for a string quartet, then the music assistant will write it. Often the composer will use assistants to orchestrate large sections of the score. What happens when the, already very developed, music AI tools can create variations on a theme and fit them to a scene?
AI can only reflect what it has learned. It will study existing material and apply what it has learned. And, as the use of AI grows, so it will be trapped in a feedback loop, learning not from real creativity but from increasing amounts of AI-generated material, and unable to evaluate objectively; ‘good or bad, it has been done before so I must add it to my set of rules’.
Great work involves understanding the rules, the structures and conventions of our industry, then turning those ideas on their heads to create something original and exciting. But how can you do that if you do not understand the fundamentals because you have never had the opportunity to do it for real?
Using technology to lower the barrier to entry in our industry is a very good thing, and I am entirely in favour. But democratisation does not mean that everyone can do something without learning how creativity really works. I do not have the answers to how we continue to develop practical talents across the industry. But I know we have to start the debate. n
Virtual production emerged as a lifeline for the entertainment industry during the pandemic, rescuing numerous shows and films from being shelved indefinitely. Its ability to create immersive worlds and enable remote collaboration became instrumental in keeping productions alive when traditional methods were constrained.
However, a persistent challenge still looms large: the industry-wide skills gap. Many aspiring filmmakers and professionals have found it exceptionally challenging to access comprehensive educational resources and courses to hone their virtual production skills. Nevertheless, a ray of hope has emerged: the UK now has what is understood to be its first full-time undergraduate degree programme in the discipline.
With established campuses in Nottingham and London, Confetti Institute of Creative Technologies runs the three-year course for an initial intake of around ten students. This follows a successful VFX course the company has had in place for the last seven years, with an average intake of between 19 and 30 students, and scoring 100 per cent student satisfaction in the National Student Survey five years in a row.
Established in 1994 by Craig Chettle MBE, a former sound engineer and tour manager who collaborated with luminaries such as Nick Cave and James Brown, Confetti identified a market need. It aimed to provide
vocationally-oriented higher education in the creative and entertainment sectors. In 2015, it joined Nottingham Trent University.
“When the industry started to talk about skill set shortages in this space and the job offers that are out there started to have more in the way of games technology and games engines on the kind of required skills list, we kind of jumped on it,” says course leader, Brian Hurst. “We thought, okay, we’re missing a beat here in terms of an undergraduate course that is dedicated to three years of teaching in virtual production which could fulfil that sort of skill set shortage.”
It evolved from Confetti’s pre-existing VFX course, specifically its Emerging Technology module, which offers a flexible curriculum from year to year. Hurst holds high expectations for virtual production course graduates, drawing confidence from the success of the VFX programme.
“We have had graduate success [from the VFX course] with students going out and immediately starting to work on virtual production content simply because they were being taught on Unreal Engine in their third year,” he continues. “One of my graduates got a job at Dupe VFX and the first project he worked on saw him use Unreal Engine to work on VFX for a Netflix feature film.”
Other success stories include alumni going into the industry to work in Scanline, Jellyfish, Framestore and Milk.
The virtual production skills gap led to a shortage of qualified and experienced professionals. Now, the UK’s ‘inaugural’ degree programme in the field aims to train tomorrow’s experts, writes Robert Shepherd
“James Cameron used virtual production years ago on Avatar, but he’s renowned for being an innovative filmmaker – it was just encapsulated as part of the way he made films – so it didn’t mean that it would necessarily catch on everywhere else,” Hurst continues. “John Favreau moved it on more recently with The Lion King, but it didn’t really get the limelight it deserved until Covid and as a result of that, The Mandalorian was pretty much all shot through virtual production. It was one of those shows that managed to do really well whilst others were shut down. I believe that for Star Trek Discovery season three, all of its VFX were created by remote artists.”
Even though virtual production was common practice for the best part of two years, Hurst says Confetti decided to bide its time before introducing the degree course.
“We spoke to some people about it, mainly in the Nottingham area and they generally thought it was a fad and that the practice would disappear into the ether along with a number of ways in which ‘you could do things’,” he adds. “Others said it’s not just another thing, it’s just a tool like colour grading, the software is a tool you can use but you don’t have to. That’s kind of true because you don’t have to adopt virtual production to create a film, but the sustainability argument can’t be ignored,” he continues. “Rather than a film crew traipsing long distances and carrying a lot of kit around a protected environment, you can now recreate that with photorealistic levels of detail using game engines like UE5 and Unity.”
That said, the real catalyst for Hurst was learning what virtual production was in a broader sense. “It wasn’t just about LED volumes, but using game engines for pre-visualising films, TV sequences and stunts,” he explains. “The good news is it involved technology we already had available to us. It was using Maya 3D software, virtual cameras and iPads as well as game engines. When I read about the teaching of virtual production within VFX, I got my VFX teams to engage in it because it made them more employable. We thought that could be a module in itself and then it grew from there.”
When Confetti was designing the course and was asked where students would come from, the team decided to make the course appeal not just to undergraduates, but people who already have a career in filmmaking. “They can come, retrain, learn to set up an LED volume; we offer that in the one-year postgraduate course,” Hurst adds.
From an initial intake of ten students back in 1994, almost 30 years later Confetti has educated more than 12,000 students, with alumni going on to work for brands such as Warner Bros, BBC and Bauer Media. Some graduates are working behind the scenes of movie hits such as Star Wars: The Rise of Skywalker
All of Confetti’s courses are conducted on-campus in Nottingham, offering students real-life work experience across its companies, which include a live entertainment venue, a fully established TV channel and a world-class esports complex, during their studies.
In September, Confetti launched a multimillion-pound site in East London, with dedicated studios and teaching spaces, and a multipurpose, 600-capacity live events venue for music, esports, virtual production and other emerging technologies.
Building on the international reputation of its Nottingham campus, the 35,000 sq ft campus in Whitechapel, is a new specialist centre
for digital arts, production and performance offering a range of undergraduate- and postgraduate-level courses in the creative industries.
“We’re very excited with the launch of Confetti London,” concludes Hurst. “It’ll be another incredible campus for those looking to forge a career in the entertainment industries, including virtual production.” n
“One of my graduates got a job at Dupe VFX and the first project he worked on saw him use Unreal Engine to work on VFX for a Netflix feature film” BRIAN HURST
Do you remember what we talked about before AI? Me neither. Overnight, artificial intelligence (particularly generative AI) has become the star of the show everywhere, and the media and entertainment industry is no exception.
Like with all new technologies, the rise of AI has been greeted with a mixture of excitement and trepidation. Since there is rising concern around the impacts of AI in our industry, adopting a healthy scepticism towards new technology can be helpful. As part of this, it’s incumbent on media tech providers to ensure AI is used ethically and responsibly in all products and applications.
Given the current interest in AI, you’d be forgiven for thinking the technology is completely new. That is
only half true. At Avid, we’ve used AI in our products for about ten years. And two years ago, we redirected Avid’s advanced research lab to explore the use of AI in media production. In other words, we and other media tech players started thinking about responsible AI use long before the current generative AI boom. ‘But how exactly are you using AI?’, I hear you ask. A great question, to which there are a hundred possible answers.
One application for AI in media tech is developing more natural user interfaces to make products more accessible. This means moving away from a user interface that depends on keyboards and mice, and towards communicating intent to applications through natural language-based descriptions and mapping that intent into automated actions.
Secondly, there’s an enormous opportunity to harness the power of AI to manage and organise media and metadata. In many companies, content is dispersed across systems using multiple naming conventions and tags. With AI, traditional search will evolve from queries on keywords to queries that return richer contextual results and recommendations. AI can play a massive role in making this content more accessible, relevant and monetisable.
Thirdly, AI can help creatives enhance the quantity and quality of their output. At Avid, we view AI as a co-pilot, working alongside humans rather than replacing them. At a basic level, AI can automate manual back-end processes, allowing creatives to focus on the big picture. Bu t it can also improve the performance of algorithms and technologies we’ve relied on for years. Consider the example of speech-to-text; this technology isn’t new but is now generating richer output thanks to AI. In this case – and many others like it – how we use technology isn’t changing, just improving.
Though the use of AI in media and entertainment holds huge promise, it also presents its fair share of challenges. For instance, how will it affect attribution? And how do we ensure that models don’t develop bias due to skewed training data?
Verifying the provenance of content is crucial to preventing a breakdown of trust in the authenticity of the content audiences consume, and it’s something we can expect to become a legal requirement. We often talk about this in a news media context, but it also extends to other forms of content, from social media posts to your favourite drama series. Audiences need to understand whether content has been created by humans, generated by AI, or manipulated by AI. So, as we invest in AI to synthesise content, we also need to invest in technologies to watermark that content. Google’s recent announcement of digital watermarking to tag AIgenerated images is an excellent example of this. Beyond transparency over content provenance, it’s also vital that media tech companies are clear about how they’re using AI, such as revealing which AI
models they’re using and how they’re being trained. This helps ensure that bias isn’t introduced into AI and that technologies are unencumbered by intellectual property or distribution rights.
Media tech providers should also develop (if they haven’t already) clear policies on responsible AI use, taking into account safety, privacy, fairness, reliability, transparency and accountability. Many of these principles aren’t new, as they’re already company policy when using open-source software. It’s about ensuring the same principles are applied to AI.
It’s impossible to predict the future of AI in media and entertainment, as we’re just starting to glimpse the possibilities. While some guardrails are needed, we shouldn’t rush to seek broad restrictions on AI usage, as the industry would risk missing out on opportunities to accelerate progress.
Looking forward, our industry needs to think beyond fear and trust that AI can be an enabler for everyone. As media tech providers, we need to adopt a ‘fail-fast succeed-fast’ mentality to maximise customer value, while also building trust by being completely transparent about how we’re using AI. n
“Media tech providers should also develop (if they haven’t already) clear policies on responsible AI use, taking into account safety, privacy, fairness, reliability, transparency and accountability”
Creativity assisted by computers has existed for decades in the film and TV industry. The visual effects (VFX) pipeline has always relied on cutting-edge technology to continuously raise the bar and deliver better content in the same amount of time. Technology naturally develops alongside the creative process to deliver the films and TV series we know and love.
For example, Christopher Nolan’s blockbuster Interstellar took 100 hours to render each frame for some scenes, using advanced black hole simulators built by theoretical physicists and animators specifically for the film. This type of simulation was so ground-breaking that it went on to be published in a scientific journal and helped advance the field of astrophysics. So, this collaboration between technology and creativity is well established.
Machine learning and AI represent the next foundational layer of technology to build on, helping automate tedious tasks and equipping creatives with more powerful tools to deliver content in the same timeframe. We believe that these AI productivity tools will be the reason to bring the industry fully to the cloud. But before
we talk about AI, we must set the stage with some context and talk about MovieLabs and their cloud vision.
MovieLabs is a non-profit technology research initiative from Paramount Pictures, Sony Pictures Entertainment, Universal Studios, Walt Disney Pictures and Television, and Warner Bros Entertainment. In 2019, it published its manifesto of a vision of the media business by 2030, broken down into ten principles.
Principle one and principle two are where we will focus: “All assets are created or ingested straight into the cloud and do not need to be moved,” and “applications come to the media respectively”. In other words, no need to download terabytes of data to work on-premise (i.e. the local computer); put media into the cloud once, then run all creative applications there.
The film and TV industry is growing fast to accommodate the demand for content. This is because the streaming services’ ecosystem represents a relatively recent addition to cinema and broadcast; the appetite for content is vast and the key is to make production faster. This is a massive scaling challenge. It is not economically viable for post houses to scale up to 500 or 1,000 VFX
artists all working on premises. The cloud, on the other hand, provides this elasticity.
However, the flexibility comes at a cost, and the vast amount of media means that the charges – particularly egress fees – are often not economically viable. A decently sized Marvel movie will have around a petabyte of data in its post phase; not an amount of data you would want to egress and therefore, egress costs need to be offset by dramatic productivity increases somewhere elsewhere in the chain. So, let’s talk about AI.
Large language models (LLMs) like ChatGPT have been generating a lot of buzz recently and represent a collection of human knowledge assembled from the internet. Moreover, these have created a shift in public discourse as more people are aware of the use of AI in our day-to-day lives. Tools like ChatGPT are called ‘transformer models’. To train these models to make them useful, the model is exposed to a large corpus of text data and learns to predict the next token in a sequence given the preceding tokens.
In layman’s terms, imagine scraping the internet for millions of blogs, articles, etc, not just for the words but the surrounding sentences and paragraphs. The model begins to understand the context, so that when you ask ChatGPT questions it can guess/infer/create a new answer one word at a time from the previous context. After pre-training, the model undergoes fine-tuning on a more specific dataset; human annotators follow a set of guidelines to create or label this data. The fine-tuning process adapts the pre-trained model to perform better on these specific tasks or to adhere to certain behavioural
Rotoscoping is one of the most time-consuming processes in the workflow, consisting of tracing around foregrounds frame by frame to separate the background, taking up to six hours of work for one second of content
guidelines. (This means ChatGPT isn’t out of line or rude when you ask it questions.)
So how can we use AI in media to good effect? Certainly, there is an expectation that it will be transformative. Rotoscoping, for instance, is one of the most time-consuming processes in the workflow. It consists of tracing around foregrounds frame by frame to separate the background; it is a labour-intensive process taking six hours of work for one second of content.
Previously, I designed technical workflows for all the major streaming services (including Disney, Netflix and Amazon) and I saw the desire to move to the cloud, and how AI could really help. People entering the VFX industry wanting to composite and be creative often have to spend a lot of time doing rotoscoping before they get on to the jobs they really want to do. In an industry that is growing faster than it can train people, this is a big problem.
The time and productivity increases brought about by this improvement in tooling flip this paradigm, completely offsetting cloud costs such as egress, empowering creatives, and allowing faster iteration cycles to deliver the best creative choices possible for each film. This will be what the industry needs to move fully to the cloud.
I’ll finish by quoting MovieLabs one final time: “We can expect many mundane or repetitive production tasks to be accomplished through artificial intelligenceenhanced tools… while it is hard to predict every use case for such tools, we expect advances to speed creatives’ work.” AI and ML tooling represents another productivity shift, where each artist is empowered to deliver at ten times the rate they were able to before. n
Artificial intelligence is a hypernym that has intelligently made everything smart. It has transformed almost all dimensions of content existence, including live streaming.
The compelling statistics of the video streaming market have witnessed tremendous growth worldwide, reaching a significant value of $455.45 billion in 2022, demonstrating that AI in live streaming is not merely a fleeting trend; it is a potent instrument that empowers companies such as Netflix, Amazon Prime, Disney+, and Apple TV to offer their subscribers a highly personalised and enhanced streaming experience.
The synergy between AI and live streaming has opened up ample space for fairness, ensuring that individuals from diverse locations and linguistic backgrounds enjoy equitable viewing opportunities.
Traditionally, live streaming was a supplementary platform for leagues and broadcasters, allowing fans to watch sport on their computers or tablets. Applications like Periscope, Meerkat, and Snapchat introduced fresh challenges by enabling individual broadcasters to stream from the stadium instantly. Simultaneously, they presented new possibilities. From a fan’s perspective, this is another avenue to experience the action up close and personal.
The FIFA World Cup 2022 witnessed AI-powered cameras autonomously tracking moving objects on the pitch, eliminating the need for manual operation aided by streaming functionalities.
Although AI and live streaming are an odd couple, the benefits come with practical applications in contemporary digital events. Wimbledon 2023 had a virtual commentator, providing analysis, anecdotes, and witty remarks. By analysing historical data, player stats, and weather, the system predicted match outcomes, adding excitement to viewers.
While real-time captioning and translation for live streams engage viewers to interact with content in their chosen language, regardless of location, realtime language transcription and translation through NLP algorithms benefit viewers who speak other languages or have hearing impairments.
The popularity of live streaming has prompted content creators and businesses to monetise it. An emerging approach involves using AI to analyse a viewer’s data and customise the experience. It includes suggesting relevant products and offering real-time purchase suggestions, increasing engagement and revenue. Additionally, AI improves ad placement, making it more effective and less intrusive for viewers.
AI-driven real-time video processing is an innovative technology that uses AI and machine learning to promptly analyse and edit video content. The AI engine meticulously catalogues a game and generates highlight reels that encapsulate epic moments and intensity.
AI automates content curation for live streams, enabling producers to swiftly access pertinent and captivating material without manual searches, ensuring streams feature high-quality content. Automated live streams have gained traction because they thoroughly eliminate the need for human intervention. This not only cuts costs but also ensures seamless execution according to schedule. Additionally, it ensures streams are filled with high-quality, pertinent, and engaging content, simultaneously on various social media platforms.
Back in the day, live streaming faced challenges with manual content moderation. However, AI integration into the live feed has enabled expeditious, effective filtering of harmful material.
Most streaming platforms use a blend of AI and human moderators to enforce policies against hate speech, violence, and misinformation while empowering users to report inappropriate content. Additionally, AI can enhance security by detecting malicious activities in real time, such as hacking attempts, allowing streamers to respond promptly to prevent damage. AI doesn’t just enhance live streaming during the event, it also brings lasting value after. Through sophisticated data analytics, AI offers insights into audience behaviour, engagement, and content performance. This empowers event managers to make ongoing improvements for future streams.
AI-powered real-time translations guarantee effective communication in international business conferences and events like the World Economic Forum 2023, enabling global participation. Similarly, during the 2023 Super Bowl, AI provided real-time highlights and personalised content, enhancing inclusivity. Predictive analytics offer insights into audience behaviour, while augmented and virtual reality, driven by AI, create interactive and immersive live streams.
The global AI market is projected to grow substantially, indicating ongoing evolution. This symbiosis between AI and live streaming revolutionises event management, promising a more inclusive future for participants worldwide. The widespread implementation of AI in events demonstrates its tangible impact. Expect to see even more innovative and inclusive live-streaming solutions with advancing AI technology. n
At IBC 2023, artificial intelligence was one of the key topics of conversation. TVBEurope speaks to four key media technology vendors for their thoughts on how AI will impact the industry, and whether they are ready to embrace it
SHOULD THE MEDIA TECHNOLOGY INDUSTRY EMBRACE, OR BE WARY OF, ARTIFICIAL INTELLIGENCE?
Chris Blandy (CB), director of strategy and business development for media and entertainment, AWS
Different forms of AI tools have been used across the media and entertainment industry for decades. Today, we’re three steps into a ten-kilometre race when it comes to generative AI. With the massive proliferation of data, the availability of highly scalable compute capacity, and the advancement of machine learning (ML) technologies over time, generative AI is finally taking shape. In particular, these ML innovations have made the capabilities of generative AI possible just within the last few years. AWS is focused on how we can use generative AI to help solve the technical challenges our media and entertainment customers are facing. Many of these challenges involve applying generative AI to areas that already use some form of AI today, such as in monetisation, upscaling, media supply chain, and search.
Steve Reynolds (SR), president, Imagine Communications
The answer is both. There is a lot of interesting work going on around how to use AI to improve the quality of forecasting, to better match audiences to inventory, and to proactively optimise performance of people, systems, and networks. But this capability must be used in the right way.
Breaches of regulatory compliance are expensive and disruptive. Bad decisions about feeds and content that lead to outages can impact millions of viewers and harm the trust in brand that most of our customers find so critical. In the end, there is a lot of human decision-making that falls under the categories of creativity and/or exception-handling, which humans are still demonstrably better at.
Rick Young (RY), senior vice president, global products, LTN Used in the right way, specifically in the work we are doing, AI can streamline, enhance and drive content creation possibilities that were never possible before. Our technology stack doesn’t dabble in the more controversial areas of artificial intelligence, like generative AI. We are focused on adding efficiency and scale for our customers so they can meet their audiences’ growing demand for live content, especially live sports.
Jenn Jarvis (JJ), product manager of editorial workflow, Ross Video Both. Artificial intelligence has been around for decades. What’s changed is the ability of AI to create original content and the relative ease of access to sophisticated language models. The technology is evolving so quickly that it’s important for both technology vendors and broadcasters to set up frameworks for safe and transparent experimentation. The capabilities are advancing faster than our ability to define use cases. It’s really exciting, but there’s still a lot to learn. If we wait, we’re losing out on important advances in productivity, but we need transparency and reflection as part of that process.
HOW DO YOU SEE YOUR COMPANY WORKING WITH AI BOTH NOW AND IN THE FUTURE?
CB: Amazon has invested heavily in the development and deployment of AI and ML for over two decades. Our goal is to help customers uncover new ways of automating processes that may be manual, outdated, and take time away from valuable creative ideation. There are four components to our approach to help customers leverage the full potential of generative AI applications. First, we know model choice is paramount, as there is no one model that will rule them all; it’s about choosing the
right model for the right use case. Then, the ability for customers to securely customise these models with their own data is key. Next, easyto-use tools are critical for democratising generative AI and increasing employee productivity. And underpinning all of this is the need to deliver responses that are low cost and low latency, which is made possible with purpose-built ML infrastructure.
SR: We’re exploring the use of AI to improve audience forecasting and revenue/yield predictions. This is a good early use of AI because it’s not a part of the real-time system. It can run early enough that there are checks and gates in the end-to-end workflow to ensure that business protections are in place.
Another area we may look at is performance and quality monitoring. One task that AI is well-proven in is pattern recognition via machine learning. If an AI system can predict network issues like bandwidth contention or quality impacts, that helps everyone run more effectively.
We’ve also seen some great work with aspects like scene recognition to enable inserting ad breaks. Especially in lower-value content – for AVoD and FAST – the ability to automate this process will rapidly expand the universe of titles and ad placement opportunities.
RY: At LTN, we are looking to leverage AI to help drive more efficient and scalable technology-powered production workflows. LTN Arc is designed to be a unique technology and services-driven solution for rights owners and rights buyers to reach audiences with versioned, relevant content on a regional, platform or even specific audience or community basis. That scale of versioning can only be achieved through technology advances driven by automation and, at times, AI.
JJ: We currently have AI built into a number of Ross products including transcription in our Streamline media asset management system, facial tracking in our Vision[AI]ry robotic camera control system, and the ability to summarise and rewrite content in our Inception newsroom system. We’ll continue to look for ways to add AI to our products to make them smarter and more efficient.
We’re even experimenting with AI in our development and technical support processes. At the end of the day, the focus isn’t on specifically using AI. The focus is on solving problems and providing innovative workflows, and AI can be a tool to achieve that.
SR: AI is a tool. To the extent that it’s a better tool – where better means more efficient, lower cost, higher quality, etc – we should try to use it. But just like any new tool, you have to find a way to measure it against the old tool. It’s a bad business decision to put something new into production without an objective comparison of some sort. AI doesn’t magically break that age-old tenant of solid business management.
RY: Automation has been a huge theme in media technology for decades now. AI and other forms of automation can be strategically implemented to make operators and production teams more efficient. And, we are hearing from our customers that this efficiency can lead
to more content creation to feed the growing demand for live event tonnage across consumer platforms of all types.
JJ: I think there will always be a place for humans in the process. Humans will be there to add nuance and perspective, to start and to end the process. But in the middle, where there are well-defined tasks and processes, machines will be able to scale faster and more efficiently. We can use AI to summarise a speech or a game, rewrite the content into specific formats for different delivery platforms and even translate it into multiple languages. But, I still think humans will need to be there to decide what stories or events have the most value to an audience, and to provide context for a local community or unique demographic. While some human jobs will likely be replaced over time, it could also open up new types of content and more extensive coverage as it makes workflow more efficient. This could actually free up many journalists and content creators to do the work they are passionate about and over time will remove a lot of the process that many of them find tedious.
CB: Like any tool, generative AI needs to be used responsibly. It will be up to the industry to agree on what the use of generative AI will look like. AWS is committed to developing generative AI services responsibly, including building foundation models with responsible AI in mind at each stage of our comprehensive development process. Amazon is actively engaged with organisations and standards bodies focused on the responsible development of next-generation AI systems including NIST, ISO, the Responsible AI Institute, and the Partnership on AI. Amazon has also signed voluntary commitments with US President Biden to foster the safe, responsible, and effective development of AI technology. We believe we can both drive innovation in AI while continuing to implement the necessary safeguards to protect our customers and consumers.
SR: This is an interesting question. There is a tendency to say that AI is just a new technology, so it should be handled the way any new technology would be handled. But that’s not the whole story. AI adds a dimension that stirs in things like consumer data, viewership info that can go as deep as packets, privacy, etc. So we have to treat it differently because it’s effectively a one-way door.
The other complexity here is that many of the operators we work with are global or super-regional. How do you align the regulation across those multiple jurisdictions and regulatory regimes? This seems like a situation where the industry may want to be proactive and agree on our own self-monitoring. Something like a voluntary industry agreement that parties can sign on to uphold. That has worked in other industry segments as an effective way to come to an agreement with governments and regulators when new technology is introduced.
JJ: I think it’s less about regulation and more about transparency. The public will decide what level of AI-generated content they are willing to accept. But it’s the role of broadcasters to be transparent with the use of artificial intelligence, and the role of technology vendors to give them tools that enable transparency. n
IBC this year was all about artificial intelligence. It was everywhere. But largely, people were talking about glamour applications, and what you could do creatively with large language models, or with generative AI applications from OpenAI like ChatGPT. But if AI was the phrase of the year at IBC, there was another theme running through everyone’s discussions. In one word: ‘scalability’.
There is a recognition that today we are making a lot more content and delivering it to more audiences over more platforms. We have finally reached the goal of ‘the content you want, when and where you want it’. K-Pop fans in Seattle and Sheffield can watch a concert in Seoul. NFL fans can watch the Dallas Cowboys from their homes in Delhi. How are we achieving this? By using the internet to carry high-quality live feeds from wherever the action is to wherever the audience is.
The challenge here is that, as we all know, the internet is not very reliable. It is an uncontrolled environment where nothing can be guaranteed. According to the tracking site thousandeyes.com, there are as many as 300 major network failures every week, close to half of them in the USA.
If you have valuable content that you need to carry over the internet, then you need to identify the best route, in terms of guaranteed bandwidth, availability and reliability, and cost. Then you have to force your signal over that chosen route, all the way from origination to destination. That, you will recognise, does not offer either reliability or scalability. There are too many variables, and the challenges escalate exponentially as you add more signals from more events. Remote production is becoming commonplace, but that means guaranteeing the feeds from 20 or more cameras, plus sound, in synchronisation, across this wild west of a network.
This is where AI and machine learning become vital. It would be impossible to build a map of the internet and identify the best paths for high-quality video streams manually, even if it was a steady state. Network failures mean that the map is constantly changing. But machine learning can identify all the paths you are likely to need and draw on historical data to build reliability models to predict how resilient each connection would be.
AI could constantly monitor every node and link in a stream, checking for noise, bandwidth and latency. When it detects a problem, it should switch to alternate routing quickly in order to provide no disturbance to the stream.
To achieve this goal calls for an integrated, multi-layer approach, one that manages both the content and its routing and protection. Modern protocols like SRT and Zixi are excellent in packaging and delivering the content, but they are at the mercy of the network. Dynamic network management can provide consistent signal delivery, but by definition sometimes it will be forced to route signals over longer distances and more nodes, thereby adding more latency. It will also inevitably have to work around signal perturbations as it detects problems and initiates a reroute. So, the signal format has to be integrated into the delivery system, incorporating a fast-forward error correction algorithm that protects the content while ensuring that switches are accomplished as quickly as possible.
At Caton, we have developed a reliable solution: Caton Media XStream. This takes signals from encoders at the venue, over the public internet to one of our points of presence, over the Caton cloud (itself an internet overlay), then back to internet connectivity for the last mile. From door-to-door, the network is managed and the system makes switching decisions in less than 20 ms.
That means the system offers six nines reliability: 99.9999 per cent up time, for live UHD and beyond. We have users running over our connections for many months with not a single outage. It can only achieve this thanks to the speed of decision making in this practical AI application. Even a single connection could not achieve this with manual management. It is the deep learning and AI decision-making that is critical.
The internet is unreliable, and we cannot change that. But AI has the power, and the intelligence, to look across many hundreds of mediocre, unreliable links to create one extremely reliable path, dynamically switched to ensure the signal always gets to its destination, error free, jitter free, with minimal delay and minimal cost. That is a practical use of AI, and it is available today. n
StoryFutures, run by the National Film & Television School and Royal Holloway, University of London, drives forward the creative and technical practice of immersive storytelling, using tools such as virtual production. Its aim is to ensure the industry thrives by developing a constant stream of new talent.
It recently published a report, identifying the skills gaps and needs across film, television, and game development. The key finding was that, despite the availability of new training initiatives, the demand for training continues to outstrip supply.
The report found that, of the creative companies surveyed, staff typically have less than six months experience of virtual production. Two-thirds reported that in the employment market, the skills required are in weak or very weak supply.
Virtual production is still a technology in its infancy, at least in practical, professional implementation. Some 79 per cent of organisations in virtual production are actively carrying out R&D on live projects, suggesting that people are, worryingly, making things up as they go along.
The conclusions are obvious. While you only get a real understanding of professional work by working professionally, there is a critical need for a strong stream of new talent entering the industry that has a good understanding of the tools that deliver immersive productions.
StoryFutures is a UK body, but it is likely that other development markets are also seeing the same critical talent gap in new media technologies. And, of course, the UK has always been a leader in media production, with talent grown here providing resources and guidance around the world.
The good news is that a growing number of UK universities have recognised this skills shortage and are developing courses which are directly relevant to the industry. To succeed, these courses have to be backed by professional-grade technology, so that students graduate with in-depth experience on the tools which they will encounter as they move into employment.
As a systems integrator, CJP has built a number of these facilities. A good example is SODA: the School of Digital Arts at Manchester Metropolitan University. This includes two greenscreen studios, a film studio and two motion capture stages. For the University of the Creative Arts we built a large LED volume with Mo-Sys camera tracking and real-time augmented
reality graphics, plus a motion capture facility with Vicon body tracking and Faceware facial mapping. And there are more.
These facilities mean that students specialising in creative production or game design get plenty of hands-on time. In terms of filling the professional skills gap, they have a big jump-start by knowing how to get the best out of the technology. But they have another important role in bringing awareness of the capabilities of virtual production and other cutting-edge techniques to other disciplines, and thereby to other industries.
Professor Arabella Plouviez, academic dean of the Faculty of Arts and Creative Industries at the University of Sunderland says of the facility CJP built there, “This is a remarkable resource for our film and television students, but what really interests me is the opportunity to involve students from across the faculty.
“It can involve writers, musicians, performers; it can involve animators or fashion students; it may bring in colleagues in business law or tourism,” she adds. “The potential is here for a whole lot of students to work together and create.”
Students working with these facilities will not only have a creative impact in the coming years, they will also be driving forward the technology, building the tools which will make virtual and immersive production viable for all.
Professor Peter Richardson of StoryFutures explains, “With so many companies in the UK poised to add virtual production to their already world-leading production expertise, the potential for the UK creative workforce is enormous. However, our research tells us that UK creative industries risk losing their R&D advantage if we fail to train the creative teams and companies who will innovate, disrupt and advance virtual production techniques and technologies.”
For all the reasons we have routinely discussed over the past few years, the demand for high-quality content continues to grow exponentially. To make material stand out it has to show creative originality, and virtual production is one way in which this inspiration can be realised.
Universities are committing themselves to developing new talent and expertise. The industry – developers of tools and production companies – must back them up by getting involved in their local colleges, supporting students, sponsoring productions and evaluating new technologies. The skills gap is real, but collaboration and co-operation will help to bridge it. n
There’s no doubt that AI and machine learning (ML) technologies are revolutionising media production and post production by enabling efficiency gains and automation across a multitude of tasks. However, reliance on public cloud services for AI-driven storage and processing is beginning to reveal some unintended consequences, and these are causing media companies to re-evaluate their strategies. Rising costs, often unpredictable egress fees, as well as lengthy upload times, are all rather unpleasant flies in the cloud storage ointment. How can media organisations utilise storage more effectively to address these challenges, yet still capitalise on the benefits of AI and ML?
Media companies today are under enormous pressure to produce more content, in less time and at a lower cost, all without compromising on quality. This is not achievable unless media workflows have an effective storage system underpinning them. Just as a solid foundation provides stability and support for a building, an effective storage infrastructure ensures that with an ever-increasing volume of data to manage, media assets are securely stored, organised, searchable and readily accessible. An effective storage system will intelligently integrate with the workflow at every stage, so that users have access to the assets they need, when they need them, from wherever they are working.
To maximise workflow productivity and avoid inefficiencies, organisations also need to focus on optimising their storage infrastructure by ensuring that media assets are stored in the right kind of storage, whether that be cloud, hybrid, or object-based. Determining factors will include organisational and workflow requirements, as well as when and how often access is needed, and for what purpose. A one-size-fits-all approach to storing media assets will most definitely not do. It is critical that all storage needs are intrinsically linked so that regardless of whether media assets are stored in nearline, archive or cold storage, they can be easily located with a unified search function.
Although media storage is the foundation of an efficient workflow, it’s nothing without metadata to enable efficient search and retrieval. AI processing services typically associated with cloud storage, provide media companies with advanced metadata capabilities for automated metadata tagging. AI and ML algorithms can enrich metadata by analysing media content and automatically generating metadata tags, making it easier to categorise, search for, and retrieve assets. This allows content to be more easily repurposed and monetised, something which may not have been cost-effective in the past.
Media workflows typically involve a lot of manual, time-consuming and repetitive tasks, making automation an essential tool for maximising workflow efficiency. AI and ML are playing an increasingly important role in enabling automation across media workflows with a wide range of applications. From metadata tagging for identifying audio and visual anomalies to AI-driven speech recognition systems for transcribing audio content, allowing for the automatic creation of subtitles, closed captions, and searchable transcripts, AI and ML allow media companies to streamline workflows, boost efficiency, reduce manual effort and enhance content quality. So, how can media organisations continue to benefit from the advantages of AI and ML tools that cloud storage offers but without the sticking points associated with the cloud?
By employing a strategy that sees AI-enabled functions being leveraged for both remote and on-premise content production, media production companies stand to gain by achieving faster processing time with fewer costs and less effort.
Localising AI-driven processes essentially means deploying AI applications in an edge computing environment. In media production, this entails carrying out AI processing tasks at the locations where the content is produced, rather than always in the cloud. Not all AI and ML functions necessitate extensive data sets to be housed in the public cloud. Many tasks are linked to smaller data sets, making them well-suited for localised AI. This approach enables media companies to craft tailored AI-driven workflows that align precisely with their unique content creation needs.
By reducing dependence on the public cloud for AI and ML tasks, organisations can substantially reduce the volume of data that is transferred in and out of cloud storage, resulting in faster processing times, and cost savings from reduced egress fees.
AI and ML technologies are redefining media production, and it’s fast becoming clear that the key to fully capitalising on the transformative potential of these technologies is to utilise storage more effectively. By shifting away from an all-in-the-cloud strategy, AI processes can be localised, and AI and ML can be adopted at the edge.
To reach the sweet spot of faster processing times with less cost and effort, content producers need to achieve a harmonious balance between cloud and edge computing, leveraging AI and ML functions for both remote and on-site content production. Only then will the true potential of AI and ML be unlocked. n
Equity and inclusion in the workplace are crucial to overcoming the challenges faced by women within tech and engineering. However, these initiatives often fall by the wayside without the right culture to support and empower them. A Women in Tech survey in 2023 showed that although more companies are addressing the gender gap in the technology sector, women still only account for around 26 per cent of people working in IT/tech. We know that the big reasons women participate in tech fields at lower rates are a lack of role models, negative peer pressure, and workplace harassment.
It’s not a question of skills or aptitude, but rather the culture that exists in too many tech companies today. We all need to make significant strides in prioritising diverse talent and supporting the belief that our competitiveness and future success are based on the need to hire and empower women trailblazers in all positions. I’m proud that we’ve started to do this at Qwilt by ensuring critical internal company initiatives include women in making key decisions.
It’s clear that more is needed from companies to actively reset the existing tech culture, including implementing initiatives that start in schools and continue on, empowering women to be role models across the industry. Recent PwC research with over 2,000 A-Level and university students showed that the gender gap in technology starts at school and continues through every stage of women’s careers. Only 27 per cent of female students surveyed say they would consider a career in technology, compared to 61 per cent of males, and just three per cent of women say a tech career is their first choice.
We must do more to encourage and make IT/tech a compelling opportunity for women seeking a meaningful and successful career, and it all starts with the right training and education. At Qwilt, we took our biggest teams and encouraged women to become leaders and role
models for their colleagues. In all roles and backgrounds, we gave managers throughout the company the essential responsibility to ensure we retain and motivate a diverse culture. That’s how you set the next step forward, establishing a culture, and driving continued change.
Research from WomenTech Network shows that at the current pace of change, it will take us more than 132 years until the economic gender gap is closed. In the US, there are over 3.7 million women working in tech positions (which represents only 23 per cent of the labour force). In Europe, there are approximately 1.7 million women working in tech positions (representing about 19.1 per cent of the ICT labour force). So, how do we accelerate these numbers in the coming years?
What we need are actionable, and concrete plans that begin with all tech and IT organisations working in tandem as an industry to promote, attract, and educate women. Qwilt is a perfect example of this thoughtful and deliberate approach, as we’ve been able to drive change at the highest levels, with over 44 per cent of our extended leadership team currently women; that’s almost half of the executive team and all the leaders who report directly to them. In addition, we’ve allowed women’s voices to flourish; women have authored 50 per cent of Qwilt’s thought leadership articles and over 60 per cent of the company’s keynote and panellist speakers were women in that same timeframe.
With time and consistent effort, these plans uplift women’s voices and contributions to make a difference in our lives today and to inspire a new generation of women-led talent. With time, we will strike the right balance and achieve our goals if, as an industry, we work together towards this critical goal. I call on my colleagues to join us; implement your own internal plans to drive change and help us deliver difference. n
DISTRIBUTED
SOFTWARE-DEFINED
FLEXIBLE
SCALABLE
Hybrid infrastructures have never been easier to achieve. As a highly dense IP gateway and processing solution, HorizoN combines the simplicity of SDI with the interoperability of ST 2110 and provides powerful UHD video processing such as SDR/HDR conversion, UDX conversion and color correction Thanks to a flexible application concept, HorizoN scales with your needs and is an indispensable appliance for any state-of-the-art video infrastructure.
