Video AI yang dibintangi Anda, dibuat dalam hitungan menit

Video transcript: I'm Mustafa, I'm the CTO at Coursera. I lead the engineering team and also help lead the AI efforts at Coursera to drive learning experiences at scale. As CTO at Coursera, I work very closely with my team to build a lot of the technology that is delivering these learning experiences at scale. We're using generative AI to do that, to really, really scale our personalization efforts, remove language barriers, and deliver learning outcomes at scale. Also, there is a lot of recommendations and personalization infrastructure that we're building because we want to truly personalize the experience for all our learners across the globe. The ability to take static content and deliver immersive content that's really, really tied to the instructor's emotions and teaching styles is only possible using videos. And now with AI and video, what you can do is translate that same video content into different native experiences. And so what that does is really helps us reach learners globally, which is what we want to do. We were not really able to translate video. We started with only text. And so being able to do that across our different content types, and video is a huge part of our library, as you can imagine. We always thought that we could use AI to translate content. We have done that with text translations. But we never thought that we could make those videos look authentic and carry the instructor's emotions and preserve them throughout the video. We always thought that translation, especially doing it for videos at scale, is going to be really, really expensive. So, you know, we would either need to invest a lot of time in building some technology that's able to do this, or we would have to hire a lot of video production services to be able to roll this out. So when the team was able to build this in a matter of, you know, few sprints or few weeks, that was only possible when we saw, when we used HeyGen and we saw that, okay, you know, this looks very natural. This looks really, really high quality and something that our instructors would be proud of, as well as our learners would be able to absorb content to drive learning outcomes. We were surprised about how fast we can, we can get to, get to this kind of scale. Surprisingly intuitive content. Quality was higher than expected, especially surprised by how it was able to conserve the emotions of the speaker and the lip sync was felt very, very natural. We saw two metrics move, which were very meaningful. One was watched video time, which increased by about 40%. And then we also saw completion rates go up by 25%. This showed us that, you know, it's not just delivering more efficient learning experiences, but it's also leading to better outcomes. HeyGen's high-quality capabilities around lip sync, voice dubbing, and doing that at scale for videos has been something that we really, really like because we are able to use that to solve for Coursera's mission, which is delivering learning experiences worldwide. The key initiative at Coursera is around AI dubbing that HeyGen has enabled and the key parts of AI dubbing that we leverage HeyGen for was voice cloning and lip sync. It was amazing to see how HeyGen is able to do that at scale as well because we were able to roll it out for a lot of different courses in a lot of different languages. The magic moment for us was when we first saw the instructor video that was dubbed with lip sync and also preserving their emotions in the video. That's when we realized that this technology can be really, really game changing for us to be able to deliver personalized learning experiences to our learners worldwide. When we started with just one video and saw the impact that this technology can have on personalizing video experiences, that really made us believe in how we can scale this. The second is this technology is moving so fast, and I feel like HeyGen is also advancing so quickly. It's hard to believe what can be possible, so it's always good to prototype a few things, try things out to see how it can fit well with your product or your experience.