2023 was the year generative AI exploded (more on that below!) As an AI dubbing provider in the space, here's what we've been up to. And if you're more of a words person (unlikely!) the written version is for you.

What happened client side:

• Questar’s Papercup AI-dubbed show My Classic Car launched on PlutoTV.
• Fuse Media launched its Spanish Papercup AI-dubbed Backstage FAST channel on Samsung TV Plus in Mexico.
• We won a Digiday Technology Award for Best AI Tool for our work with Fremantle which has resulted in 1 million views a month on its Spanish YouTube channel and increased watch time on Fremantle's Arabic channel by 300%.
• Insider’s AI-dubbed content hit 910 million views across its socials.

What the machine learning team worked on:

• Making huge improvements in the expressivity of our voices, so that they’re as realistic as the voices we hear every day – in conversation, shows, and movies.
• Deploying our human-in-the-loop highly expressive voices at scale (and delivering our largest FAST deals with it. (What is voice conversion? A process in which a new synthetic voice is created by modifying the characteristics of the original voice.)

What happened on the product side?

• Utilizing large language models to ideate and build user experiences that haven’t previously been possible. With large language models, we’ve been able to get better context, use multi-modal models to analyze images and provide our localization team with all of the assets they need to build the highest-quality videos. 

What happened beyond Papercup

2023 was the year people with internet access fully integrated generative AI into their daily lives,  mainly via Open AI’s ChatGPT. Now used by 180 million people, it’s as if it always existed. In the year since its launch in November 2022, the chatbot has become the fastest-growing consumer software application in history with a valuation of $29 billion. 

2023 was also the year of eye-watering living costs and attendant frugal consumer behavior. Against this backdrop, ChatGPT became a symbol of a wider trend: individuals and companies laser-focused on efficiency. 

Cord-cutting reached an all-time high at the start of the year, falling to its lowest level since 1992 according to Variety. In tandem, consumers became more savvy about how and where they consumed premium content with a rise in subscription hopping – whereby people subscribe, watch a series, and leave again. The fallout of high costs and cash-strapped consumers was a ubiquitous bid by the biggest media and entertainment companies to reduce overheads. Staff at Buzzfeed, Disney, Insider, Spotify, Vice, and Vox were among the casualties. 

In this climate, Generative AI’s promise of increased efficiency, to some people’s minds at the expense of humans, became one of the driving forces behind the 118-day actors’ strikes in the US. 

As a dubbing provider that localizes video faster, more cost-effectively, and without integrating complicated workflows, we continued on our mission to make videos available in any language, so media companies can reach the global audiences crucial to their success. Complying with the trend, we had to become more efficient ourselves – utilizing large language models to ensure our processes were lean and optimized for the best possible output. We responded to the needs of the media companies we work with by making our voices more expressive so that they meet the quality expectations of audiences. Here are some things we learned along the way. In the context of strikes, cash-strapped consumers, and media companies striving to survive and thrive, we also thought long and hard about how to balance sustaining the company, serving our clients, and paying voice actors fairly. More on that in the New Year!

Our key takeaways from the year

"The launch of automated models has shown clear improvement in underlying large language models technology, though automated output still cannot tackle the last mile of quality which is indispensable for anyone that monetizes their content." Jesse Shemen, CEO

"The FAST world has continued to mature, with major studios like Warner Brothers Discovery jumping in and content offerings getting more and more premium. Instead of competing head-on with the majors, we're seeing clever distributors look for the gaps in the slate. We've entered the era of single IP and niche channels on FAST that deeply engage fandoms."Garrett Goodman, VP of sales 

"This year YouTube introduced its multi-language audio feature which lets content owners add dubbed audio to their videos so viewers can watch them in their preferred language. The feature enables content owners to grow international audiences, boosting the channel’s overall performance. It helps creators expand their global reach and find new audiences for their channels.” Gemma O Kane, Client lead

"The industry has accelerated its adoption of alternative strategies to maintain content flow and appeal to diverse audiences globally. Among these strategies is the adoption of innovative technologies that enable the customization of content for viewers across different regions. This is a visionary step towards making all types of content more accessible to a global audience – a trend that is set to only continue in 2024." Abhirukt Sapru, SVP of commercial

"This year saw great improvements in the voice cloning abilities of technology, which attracted a lot of attention. In our experience, voice cloning alone is not sufficient – end viewers will disengage if the expressivity of the voice is not up to par. Expect improvements in AI dubbing expressivity to be the improvement that yields engagement for end-viewers and ROI for content owners." Kilian Butler, Head of product

If you want to start the new year with a bang, get in touch with us and see how we can bring you global localization that only Santa can match.