Cova, Kobe, Kenya Valley discussed on The Voicebot Podcast


And when i say commodities really looking at commoditisation on the horizon i think Epic games recently announced a human creator which is going to be free and out this year so anybody will be able to make one of these really awesome but a real faces. I we're still like you know one to two years away from synthetic voices getting better in the next one to two years everybody will have access to the technology to make a great looking face or you know the the virtual influence the static image on instagram. That doesn't talk There are companies. You know who started doing that. For years ago. Where production costs exorbitantly high. And they were claiming they were a virtual characters but the religious posing real models photo shopping. Cgi had on top. And the technology's really just not interesting. So i don't. I don't think that it's a fatal flaw. But i think that people who are working on that side of the spectrum are underestimating. How far away the conversational. Ai pieces and i've heard so many people let go like waiting for g. P. t. three access and like it's it's all going to work out and then like you know couple of people who are like. Oh we tried. Gpc three it's really slow off the rails doesn't even know about cova because it was trained before kobe. All all of these various things but zona fatal flaw for avatar technology. Because i think like with epic games matt humans the use case that they have in mind far future yes conversational a one hundred percent on their radar but today like i think they're more interested in moving obviously first and foremost core business game character production and those are all scripted. Pc's as part of the game and then also more real production so these characters being used in movies media entertainment so there's still going to be a huge market for digital production which doesn't mean conversational ai to be successful but for people who are marketing that they have like an a character virtual character that can scale to you know. Talk to really connect one on one with an audience at once are not going to be able to deliver on that vision With the tools that are available today we know very few exceptions which if there are companies out there like ours have data and half working on this for a long time. We'll have been building discreet. Unspecific character. And so i. I could see that. I think that makes a lotta sense. So you mentioned meta humans from unreal engine. Are you thinking about either using that for kooky or maybe another character that you would develop. They would have the the human like characteristics or are you going to stick with the the more illustrated format so in terms of developing her new avatar her latest. Look we ran extensive user testing with our existing fanbase. We ended up collecting ten thousand data. Points actually more than ten thousand And we were testing to see if people had a preference along a range of spectrum so we tested gender like male female we tested an age range. Although for cookie the challenges like she's always been gender female like she's always been like for eighteen so she kind of has the back story that people are used to and we tested. Do people want like stylized versus photo real or do people even like non-human looking characters. Like they want the. I look like an actual robot and overwhelmingly what we found was a preference for you know female gender and actually a preference for stylized not photo real knees were static images not anything that was creepy with beyond kenya valley. Like all the tests we did. Were settling people setting out what they wanted to wanted her to look like while they were chatting and so that was very very.

Coming up next