Continuing the trend! Here are some stories from the past week in gaming, machine learning, and audio which I found interesting to share and discuss.
There’s been a good amount of experimentation recently with “interactive media” such as Black Mirror’s Bandersnatch. But for a while now, there’s been a community with a different take on what interactive entertainment means – and that’s streaming. On platforms like Twitch, instead of pre-baked content that’s wired to respond to a viewer’s input (and therefore behave differently for each viewer), you instead have a real person creating content live, and engaging with their entire audience – creating a unified and dynamic experience that fosters a greater sense of community among viewers. Streaming so far has largely focused on creating a new kind of content – streamed gameplay – but it only makes sense that they’d explore other kinds of content as well, especially as content such as “Just Chatting” channels take up an increasing percentage of viewership. Twitch seems unlikely to create a single blockbuster show, so I expect their biggest focus will simply be enabling many streamers to create new content as quickly and effortlessly as possible, in order to capture the same size audience overall.
Not much to say about this – just a really impressive tech demo showcasing the upcoming developments, especially in dynamic visual rendering. Excited to see these improvements begin showing up in games as the new engine and console are made available!
At first glance, I must confess that I had no idea why this strategy would be a good way to earn revenue. Xaviant appears to be incentivizing its players against playing more, meaning they’ll have difficulty developing an attachment to the game or the community around it.
That said, as I’ve read more, it’s become apparent that I may be optimizing for the wrong thing. Xaviant’s blog post mentions the expenses of running the service, so it’s possible they chose this strategy not as ideal for revenue, but instead aiming to minimize their downside risk. If so, perhaps this speaks to a need in the gaming space for a cheaper way for young games to build a following before having to worry about infrastructure costs? Which leads us into the next bullet point…
Epic has released a wide range of online services, including things like matchmaking, analytics, and even its player network graph, to be used for free by any game developer. It’s missing some of the costlier infrastructure components (such as directly hosting chat or game servers, as best as I can tell) but still a huge step in democratizing game development. Of course, this is also in line with Epic’s mission of creating a “multiverse” around their products – using their online services or other tools like Unreal automatically means building your game on an interface that’s easily compatible with Epic’s overall ecosystem, enabling them to more effectively aggregate content onto their platform.
There’s no doubt Epic has a powerful distribution channel through Fortnite’s ever-expanding player base, but gaming is a big space, and there are many other participants who are still competing effectively with Fortnite for gamers’s attention. So it will be interesting to see how Epic balances their focus on enabling content creation for all, with their need to continue being the ones at the forefront of innovation relative to other would-be platform leaders.
Both research organizations focus on longstanding problems which have the potential to unlock text-to-speech for more entertainment-focused applications. First, Facebook’s research focuses on reducing the compute cost of text-to-speech, ultimately getting it to a place where text-to-speech lines could be smoothly generated on the fly in an interactive context. Their current solution required a lot of custom work and is still pretty heavyweight (relying on multiple CPU cores for concurrent computation), but it’s a big step in the right direction, especially in terms of enabling text-to-speech in systems which may lack custom neural hardware.
On the other side, Nvidia has been focusing on making more emotive text-to-speech, and in particular developing more reliable “levers” by which one can actually tell their text-to-speech system which emotion it should be expressing at a given point in time. This lack of customizability has been a big problem for expressive text-to-speech so far, and success here opens up a broader possibility of text-to-speech lines for e.g. NPCs in a game which still feel authentic and emotive to the player
In response to growing concerns about the toxic trends in gaming communities, Twitch has assembled a new council to explore ways it can reshape the platform to encourage safer and healthier communities. Twitch is absolutely not the only company in gaming to be investing in this kind of effort, but I have to say I was impressed by the diversity of the individuals they’ve assembled for this team. I mean that not just in terms of things like age, gender, and race diversity (though that’s extremely important, and well represented, as well!), but also in terms of diversity of skillset and focus. The council includes multiple streamers, leaders of strategic outfits, and academic researchers, ensuring a good balance of both anecdotal/personal understanding of toxicity, as well as broader statistical understanding of large-scale effects. This sort of group has the potential for some really creative and meaningful change, and I’ll be keeping a close eye on the ideas they come up with.
That’s all for this week! I’ll be seeing how it goes to keep this up for a while, but please let me know if you find these posts useful (or not!), as any feedback is always welcome!