Pod Insights Eye On A.I. #265 Jeff Lu: How Akool Uses Gen AI to Make Live AI Avatars & Voices About this episode What if you could translate any video into any language—instantly—and make it look like the speaker was really speaking it?
In this episode of Eye on AI, host Craig Smith sits down with Jeff Liu, founder and CEO of Akool, to explore how AI avatars and real-time video translation are eliminating global language barriers. With a background at Apple, Google, and Stanford, Jeff is leading Akool to the frontier of generative AI: cloning faces, voices, and emotions in live video streams.
We dive into the technical architecture behind Akool's real-time avatars, the role of large language models in translation latency, and what the future looks like when the need to "learn languages" may disappear.
Whether you're a content creator, marketer, technologist, or just curious about what's next in AI and communication, this episode is a must-listen.
Stay Updated:
Craig Smith on X:https://x.com/craigss
Eye on A. I. on X: https://x.com/EyeOn_AI
(00:00) Creating Digital Clones with AI
(01:20) Jeff Liu's Journey from Big Tech to Startup Founder
(02:54) What Akool Does and Who It's For
(07:11) Inside the Live AI Avatar Suite
(10:32) How Akool Powers Real-Time AI Avatars
(16:05) Solving Language Translation with Avatars
(21:40) Translating YouTube Videos with One Click
(28:36) The Technology Behind Real-Time Translation
(33:22) Competing with Tech Giants Like Google
(36:26) Akool's Vision
(39:36) Avatar Types: Instant, Studio, and Ultra explained
(43:39) Building Emotional Nuance into Voice and Video
Transcript Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.
0:05 Speaker 00 You can create a digital clone of yourself and use that as a video feed into any softwares like meeting softwares, podcast softwares, streaming and so on. Our current translation is a separate feature that they do translate and animate to a real video feed, not the avatar feed. So it's actually easier for us to make the avatar do the translation rather than the real video feed. The barriers of language will be pretty
0:46 Speaker 00 trade reduced in three, 25 years. Lots of the video translations will happen, lots of the real time translations will happen. Maybe in the next generation, they don't visualize so many language anymore.
1:04 Craig S. Smith? So Jeff, can you start by introducing yourself? Give us some of your background, where you went to school, what you studied, what you were doing before this, and then tell us about Akool.
1:18 Speaker 00 Sure. My name is Jeff and my background is mainly around the engineering side. So I spent more than a decade working on the AI videos. So previously, I did my PhD in University of Illinois at Barnum and champagne with one of the big name professors over there and did a research in Stanford, working on video generation. And after that, I went to Apple, mainly working on the Face ID, working on how to generate.
2:06 Speaker 00 Understanding the face and identities and so on. And I went to Google Cloud, uh, working on video processing on the cloud, uh, especially like understanding the videos, what's happening over there and so on. And, uh, this is the first company that I started about three years ago. And I, uh, I have some other startup experience with some of the friends company.
2:42 Craig S. Smith? Yeah, and what, what is, uh, Akool's, uh, premise? What's the product? Uh, and, uh, and then I'll ask you about the tech behind it.
2:54 Speaker 00 Sure, so, um, the company was founded three years ago. And what we, uh, work on is the mainline AI videos, especially around humans in the videos. And, uh, for example, we do AI avatars either, uh, for training education, which create videos or for like, uh, live streaming and acting for meetings and so on. And well, and marketing, the influencers, yeah, lots of the stuff around avatars. And another thing is video translation. We do lots of the video translation stuff.
3:37 Speaker 00 You can, uh, give a YouTube link and you can translate the YouTube video into any other language.
3:44 Speaker 02 Hi, 我叫尤金，这是EcoLife Camera的演示。 这是我们EcoLife套件中的最新产品， 也是全球首批实时视频生成解决方案之一。 Hola, mi nombre es Eugene y esta es una demostración de Equal Life Camera, el último producto de nuestra suite EcoLife, una de las primeras soluciones del mundo para la generación de video en tiempo real.
4:06 Speaker 00 And also other things, including uh like face swap. We can uh change the face and appearance and the age of the person in the video. Um there are lots more. Our main customer base are more on uh business side.
4:25 Craig S. Smith? I see.
4:26 Speaker 00 Um studios, agencies, marketing, tech companies and so on.
4:31 Craig S. Smith? Yep. How many of these, I mean, you mentioned a suite of products there. How many have been launched uh to date?
4:38 Speaker 00 Yeah, so we already launched many products. I think we have like close to ten products on our website that's ready to use. Most of them are a video creation site.
4:53 Craig S. Smith? Yeah, and uh does that include, uh you were saying you can translate YouTube videos into other languages? Does that include uh lip syncing the videos?
5:05 Speaker 00 Yes, that include lip sync the video and well. So uh especially if they're the person like do a podcast with it kind of things. We can do uh lip sync and we can also do more complex uh situations, include moving as well.
5:21 Craig S. Smith? Yeah, well, on the lip syncing, I'm curious because uh the translate, you know, I've done, I had a company called Lip Dub on the podcast and they do lip syncing of of dubbed videos or or you can upload uh Chinese, for example, uh audio,
5:44 Craig S. Smith? Chinese translated audio to a video in which I'm speaking English. But the uh the length of the sentences differs, so you have to do some adjustments. Uh maybe make this Chinese sentence longer or shorter in order to match the video and then they do the lip sync. How do you handle that with with your uh your YouTube uh translation?
6:17 Speaker 00 Yes, so that's a good question. We provide two options. The first option is keep the original video lens. In that case, we will uh generate the sentence that's the same lens. Uh when you speak it out in different languages. And another option is called dynamic video lens. So we will uh change the video a little bit, either speed up a little bit or slow down a little bit. So you don't necessarily need to keep the lens to be exactly same.
6:53 Craig S. Smith? Oh, that's interesting. Yeah, uh well, I'll have to try that. Uh but the but the product that that we're talking about today is this live of AI avatar. Uh can you describe what that product is?
7:11 Speaker 00 Sure. So yeah, the products we are launching is uh live uh suites. And it includes a series of products and the main one is live AI avatar. Now what it does is you can create a digital clone of yourself. And they use that as a video feed into any uh softwares like meeting softwares. Uh podcast softwares, streaming and so on.
7:48 Speaker 00 Uh and we have additional offerings in the whole suite as well, including uh real time live video translations. Real time live face swap and we also have an upcoming uh real time live AI video generation.
8:05 Craig S. Smith? Yeah, now on the uh live avatar, I understand training an avatar. But how do you have the avatar lip sync?
8:18 Craig S. Smith? With the voice in streaming video, live streaming video, to me, that that sounds impossible.
8:27 Speaker 00 Yes, it's a very hard task to lip sync in real time into the meetings with a voice. And we did a lot of optimization to ensure both a quality and a speed for the avatars in the lip sync. So actually, we have a series of models that runs on different computation resources. On the high end, our models run on like G H two hundred or H two hundred GPUs. And it goes to like A one hundred and A tens.
9:11 Speaker 00 And our most optimized model even runs on the laptop GPUs. So on the laptop side, we are working mainly with AMD. And for the live avatar side, we are running on various types of GPUs based on their situation and needs. And they've been working very well.
9:35 Craig S. Smith? Yeah, so for people, because this is both an audio and video podcast, for people that are listening only, I'd encourage you to go to YouTube and watch the YouTube version of the podcast. Because Jeff, you've been, what's on screen is an avatar. Can you switch to your live view, your real self? And then we can talk more about how that avatar is is working in live streaming.
10:12 Speaker 00 Sure.
10:13 Craig S. Smith? There we go. So can you just walk through the architecture? First of all, is this proprietary, the models that you're running in the background, or is it? Is it an architecture of API calls? Or how how is it happening in the background?
10:32 Speaker 00 Yeah, yeah. So we developed the model ourselves, the whole stack. From L wisdom to the data collection, training and deployment, data book, data ability. So all of them myself developed. And we deploy these models on the cloud. And depend on different situation. Um, some of sometimes the model is we large compute such as.
11:02 Speaker 00 And sometimes they just smaller compute. Really depend on the need. And how it works is on our desktop site, you install a small. Um, a small install package, the Open H. And uh, it will connect your camera with our cloud servers. And your camera camera stream will be streamed to the cloud servers. And we want to compute on the cloud and that stream back. To the uh, service that you are using such as uh, Google meeting room or the podcast software.
11:41 Craig S. Smith? Yeah, uh, and and so the the video streams to the cloud. And the cloud streams back the avatar matched to the voice. Is that right?
11:52 Speaker 00 That's right. So uh, there. Blipsing and avatar animation, the thing happens on the cloud and stream back to the. Um, to the uh, presenting software.
12:06 Craig S. Smith? Yeah. And do you need a a other, I mean, you download this package. But do you need a special software? Uh, on your, I mean, um, special hardware, a laptop with special hardware? Or does this work on uh, consumer laptops?
12:26 Speaker 00 Um, it's consumer laptop, so it and the software. I just do very minimum thing. On laptop, it only provide user interface for you to make action. And well, and uh, take your video stream from the camera and the streaming to the cloud. So that there no compute happening on.
12:49 Speaker 00 Of your laptop or Mac. And so it's a very minimum. Compute or software requirement.
12:59 Craig S. Smith? And how do you deal with the latency? I I noticed with the Avatar, you were speaking a little more slowly, a little more deliberately. Is that because you need? To allow the for the latency streaming to the cloud and back?
13:19 Speaker 00 Yeah, so for most of the live features, we optimize the latency a lot. So the thing that we noticed that for the live interactions, you are a latency below. Half a second cannot be perceived. And it below two seconds, then it's fine. And for the for the features we are working on, we can show and optimize the latency to be one to two seconds. And in the best case, you can go down to at low at half a second, but most of the most of the time is one to two seconds latency. So that that's very acceptable.
14:04 Speaker 00 And there are lots of things that we can do to optimize latency and get into more comfortable real time experience.
14:15 Craig S. Smith? And but when you're speaking, using the Avatar. Do you on do you speak more slowly or is that?
14:26 Craig S. Smith? An artifact of the of the?
14:30 Craig S. Smith? Of the of the system?
14:32 Speaker 00 Um, I think that maybe I speak a little bit slower, so there there should be no artifact. For the voice, because the voice is almost unmodified with those. So the main thing I will modify in the Avatar is the thing. So it is. I don't think the voice will be slowed or so on.
15:00 Craig S. Smith? Yeah, can you switch to another Avatar? Just so we can we can see how it looks.
15:10 Craig S. Smith? Again, now that we understand a little bit more what we're looking at.
15:14 Speaker 00 Sure, sure. Yeah, let me use another avatar to join.
15:23 Craig S. Smith? Yeah, that definitely does not look like you. Uh, okay, so this avatar, is it one that you trained? Is it an off-the-shelf avatar? What is it?
15:36 Speaker 00 Yeah, it's a off-the-shelf avatar that we provide on the website. So I think everyone can use it. And yeah, it's off-the-shelf option.
15:50 Craig S. Smith? Uh, and you said that you're eventually going to have live uh, translation, uh, audio translation married or merged with the avatar, is that right?
16:05 Speaker 00 Yes, that's right. So our current uh, translation is a separate feature that they do translate and animate your real video feed, not the avatar feed. So uh, it's actually easier for us to make the avatar do the translation uh, rather than the real video feed. We'll make it happen.
16:31 Craig S. Smith? Yeah, well, that's it's easier because uh, it's easier to sync the lips of the avatar than it would be a live video, is that right?
16:45 Speaker 00 That's right, that's right. So sync the lips of avatar is uh, easier than sync the lips of a live video, especially if there are the random people show up in the video feed and we need to sync the lips perfectly. So that's uh, that's definitely a harder task than you already have an avatar with you.
17:09 Craig S. Smith? Yeah, uh, and and this, you know, there's, things are moving so quickly. Uh, Google just announced, um.
17:19 Craig S. Smith? Live audio translation, or I don't know if it's live translation, but where you can switch between languages while streaming, I believe. Is this a problem that's now been solved? Are you using any any other models to do this? Or can you talk about the tech behind the
17:54 Craig S. Smith? streaming translation?
17:56 Speaker 00 Sure, yeah, I'm happy to talk about that. So the streaming translation is not a fully mature area. And just only just for the voice side only. So if you think about it, there are so many languages out there and how much latency you get. It's a very challenging task, especially if you want to do real time translation. So you need to balance between latency and accuracy. The more of a sentence your voice model hears before it translates, the more accurate the translation will be.
18:41 Speaker 00 So definitely if more predictions in the AIs are being used, then less latency and more accuracy. And for our case, for the translation part, we are integrating with Microsoft to do the translation. And also there are some other players we are working with as well. And for the other piece, which is lip syncing and that movement, that's also very challenging and that's our core expertise.
19:18 Speaker 00 For their line translation, which makes you able to feel like you are speaking in in another language.
19:27 Craig S. Smith? Yeah, I mean, that's it's remarkable because again, as I say, I've been translating content from English in the Chinese video into from English in the Chinese. And the process I'm going through is I go to eleven labs. I I translate the audio into Chinese, for example. But I have to adjust the sentence lengths. And then I export those audio files, go into Liptub, train the the Liptub on the on the video so that it's
20:11 Craig S. Smith? it it can reproduce the lower part of the face and the lips. And then upload the audio and then render and or generate the new video. And it's it's a fairly cumbersome process. But you're talking about doing that from a YouTube video. Do you do you just put the URL of the YouTube video into your system or or do you have to upload the video to your system directly? And then train the system on the speakers? And I mean, how does all of that work? And why don't we switch back to your live view?
20:56 Craig S. Smith? Just it's a little creepy talking this way.
21:04 Speaker 00 Sure, yeah, let's squeeze it back.
21:06 Craig S. Smith? But but do you understand my question? How how? I mean, it just amazes me the computation required. First of all.
21:18 Craig S. Smith? If it's a YouTube video, you need to train your model on the face in the video. Then you need to translate the audio and then you need to regenerate the face. So the lip syncs with the audio. Is that right?
21:40 Speaker 00 So actually, we have a much simpler pipeline and the idea works well. So our product is all for easier usage. So if you want to translate a YouTube video, you only need to copy the URL of the YouTube video and paste it to the box. And they choose the language you want to translate to and click translate. And all the magic happens in the backend. So we do everything, including downloading the YouTube video. And so we have a model that don't need to train per person. It's a generic model. Any person comes in, it immediately works. So no training is needed, actually.
22:25 Speaker 00 And so we can do lip sync and we chrome the voice. And every magic happens behind the hood. It really works as well. And if you want to have more fun tunes of the translation, you can click into the script. Either before translation or after translation, click into the scripts and that makes up happy. And we can generate another translation version that matches your updates or modification of the script. But I think for most of the case, the translation is very good. So we did lots of the translation of YouTube videos.
23:10 Speaker 00 And the translation is, I think it's almost identical to the default YouTube subtitle translation. So it's almost same. So that I think it's already pretty good translation quite even without any modification.
23:27 Craig S. Smith? Yeah. What fascinates me about this is it's not geography that separates people. It's language. And you know, I spent a lot of my life in China. I'm assuming you're from China. And the...
23:47 Craig S. Smith? The ability to share content or view content across languages. Uh, that barrier is disappearing and I, I think that's uh. A tremendous benefit to humanity because there's greater understanding of culture and greater understanding of. Attitudes and opinions. Uh, when you do this to YouTube videos, are you then uploading them to somewhere in China? Is there a site that, that is comparable my videos, for example, I upload to Bilibili?
24:27 Speaker 00 Yeah, yeah, for my case, it's um, so uh, my wife is from South Korea. Oh, she speaks Korean and her family speaks Korean and don't understand Korean. And there was so, uh, we have uh, family members from under place to speak other languages as well. So it's uh, very meaningful for me to translate the content and understand them like. Some other videos my wife is watching, I have no idea what, what it is. But I can translate it and I can watch it. So I have to consume lots of the videos my, these days.
25:08 Speaker 00 So uh, and I just uh, see some uh, good YouTube videos, maybe in another language, especially Korea when I'm doing. I just translated and I watch it myself. I didn't post it somewhere, so I just watch it myself. And uh, we do have some of the videos. Um, trying to, trying to share in like another language on socials and so on. I've already done going and I think more on the company side, uh, not on my personal side. So on my personal side, most of the videos I translated, I just watch it myself.
25:46 Craig S. Smith? Yeah, what, what is the most popular uh, video platform in China today? For, for this kind of content?
25:57 Speaker 00 Yeah, so the uh, the comparison to YouTube. Think for the long, long form, it's Bilibili, it's very popular. And for uh, for a short version.
26:12 Speaker 00 A TikTok, right?
26:13 Craig S. Smith? Yeah.
26:14 Speaker 00 TikTok is the most popular for short video.
26:17 Craig S. Smith? Yeah.
26:17 Speaker 00 And I actually watch YouTube much more, so I I I mainly just use YouTube. YouTube has everything.
26:24 Craig S. Smith? Yeah. This tech. When did? Because I think I told you when we spoke earlier. I did a podcast in maybe two thousand eighteen with a researcher at Baidu and he was trying to do a speech to text translation. Uh, because uh, there was such a big lag time in speech to text, you know?
26:58 Craig S. Smith? The problem that he was dealing with is Baidu has a lot of English speaking uh, employees that don't speak Chinese and their executives would at a conference would be speaking and there would be uh, you know, translated uh, text. Uh, you know, on the side or on the video. But the translated text was always two or three sentences behind the speaker. So if the speaker told a joke, the Chinese audience would laugh. And then the English speakers would have to wait watching the text to get the joke and by then, you know, the speaker had moved on.
27:43 Craig S. Smith? And and the problem that this researcher was having is in the uh, prediction. It was just taking uh, too much time to predict the next word uh, to to close the latency gap. And particularly between uh, a language like uh, English and German. We spoke about that with with the the sentence structure being entirely different. Uh, when did that hurdle? Uh, when was that crossed? When there must have been, is that purely um, uh, because of the transformer algorithm? How did uh,
28:28 Craig S. Smith? the the technology uh, get over that hurdle of of the latency in the prediction?
28:36 Speaker 00 Yes, that's a very good question. So it's always hard to battle with the latency in translation. Especially all the languages are different, right? Like Korean is then structured very different. English, Chinese and so on. Maybe for some language, you need to listen to until the end before you know what the people actually mean. So it depends on their language, actually. So for some of the language, the latency can already be optimized very low. And also, it's a battle between accuracy and the latency.
29:20 Speaker 00 And also, if the people speaks a short sentence more, and latency can be even lower. But the biggest improvement in this space is indeed the transformers and the edge language models and so on. They are much better at predicting and the translation than translation accuracy and reduce latency. So and for the more specific details, and onto the expert here, we are actually using some of the third-party services for the live real-time translation.
30:05 Speaker 00 And actually, how to convert voice to language, from one language to another. So we are integrating some third-party. They definitely did a lot of optimizations to reduce the latency for the prediction and so on. They also provide us parameters to balance latency and accuracy. There's something that we can play with.
30:31 Craig S. Smith? Yeah, and so does the latency change depending on the language pair? I mean, for example, Chinese and English are very similar in sentence structure. But as you said, Korean and Chinese or German and Chinese, it's quite different.
30:51 Speaker 00 Yeah, I think it depends on the language you pay as well. So Chinese and English, the words order pretty similar. And I think for Korean, they made in German. The verb is usually at the end of the sentence. So it's a different structure. So what appear is highly related to the like.
31:14 Craig S. Smith? I see. So, so in, in English to German, uh, the latency would be, uh, higher. And, and so the video, if I were doing a live video, uh, there would be a lag.
31:31 Speaker 00 Yeah, that's right. So basically, uh, for some other video, you need to listen more before you can translate more accurate. You can also forthly to translate, but in that case, then the translation will be not accurate. Uh, but in general, we will already optimize the latency to be, uh, pretty low. So, but definitely more and more optimization in this and also language.
31:58 Craig S. Smith? So, uh, Akool is the, is the name of the company or the name of the product?
32:04 Speaker 00 So it's the name of the company.
32:07 Craig S. Smith? And how big is your team now?
32:10 Speaker 00 Yep, uh, we have, uh, seventy, eighty people around. And, uh, most of them are engineers.
32:20 Craig S. Smith? I see. And, uh, who are your, uh, your backers?
32:25 Speaker 00 Yeah, the, the company, the mainland bootstrap, we do have some, uh, investors. But, uh, I think mainland bootstrap and, and the business is, uh, uh, I think it's in, in a place that we have pretty strong product market.
32:43 Craig S. Smith? Yeah, yeah, absolutely. Uh, are, are you concerned? Because as I said, Google has come out with, uh, you know, the ability to, uh,
32:57 Craig S. Smith? Translate, switch between languages in live translation. Which sounds similar to what you guys are doing.
33:09 Craig S. Smith? Are you concerned about building market share with the the big guys looking at the same space?
33:22 Speaker 00 So I think our product is not the same.
33:27 Speaker 00 Based on my understanding, what they did is they they do a voiceover on top of the original. Video for the translation, we are not voiceover. We actually added the video track and also added the video in real time. To do the translation, so our our solution is actually. Better than just doing a voiceover on top of that. And also, Google, usually many of the players, when they introduce. A solution, they bring it into the. Into the Google suite.
34:08 Speaker 00 Right, we have a much more generic solution that you can integrate into any software. So. And also, it's actually a good thing that when there are more players over there. It means this field do. At of their. Needs and market and well as. We we can get much more users easily.
34:37 Craig S. Smith? Yeah, and how are you pricing this? Do you have a? A freemium model? Is it based on usage and that you can use a certain amount for free and then you have to pay? I mean, how do you? How do you do that?
34:53 Speaker 00 So the uh the quality I mean currently in beta. So for this it all version, it's provided for free, but with limited access. You need a flight to get access. And for the G A launch, it will be a subscription model. And there you can uh free a try it for an hour or so. And then often ads you need to subscribe to plans and then you can uh do they for example thirty hours a month for them. For the either translation, webinars or more advanced features.
35:37 Craig S. Smith? Yeah, and so that's for the entire suite of products or or can you? For example, I'm very interested in the YouTube uh translation product. I have less need for the live avatar product. You break them out or is it bundled in a single subscription?
35:57 Speaker 00 So all of them are bundled in a single subscription that we have. So we actually have a suite of tools and offerings and all of them are being bundled. So which is which is a big advantage that we have of them.
36:14 Craig S. Smith? Yeah, who are the customers so far?
36:17 Speaker 00 Yeah, yeah, so we are a platform that target more professional use and business use from the very beginning. So in the early days, we have lots of the users in the marketing phase. We have many marketing agencies and we have lots of the brand and marketing departments on the platform. And then we go more on to the uh studio and the creators and the influencers. And then we have lots of technology company on the platform and using it. So I can give some examples.
36:59 Speaker 00 Our first large client, say the Coca Cola, and they use us to run uh marketing campaigns and so on. And now we work very closely with tech companies like A W S, A M B.
37:15 Speaker 00 Helps them to provide solutions and so on. So definitely, uh, we are across a different vertical.
37:24 Craig S. Smith? Do you think eventually your technology will be embedded in uh in teleconferencing software so that the the end user doesn't have to subscribe that it'll be an option in whatever teleconferencing software you use that you can, you know, choose an avatar and a language
37:50 Craig S. Smith? and and that it takes place uh on the side of the teleconferencing software?
37:58 Speaker 00 Yes, that's a very good question. So what we believe is first, we need to reduce the compute costs for lots of things that we do. It's one of the efforts we've been working on. So if we can make most of the compute happen locally on the laptop, then I think lots of the software, they will be able to provide that and a free feature on the on the software. If still a lot of compute are needed, I think it will it will more likely to be a paid feature.
38:38 Speaker 00 And based on our understanding, most of the providers on this market, they are more focused on the voice feed. We didn't see too much players are working on the video feed or the imaging, more on the on the voice side. So maybe voice will become um the fourth option in the near future, like probably like two or three years or something. On the video side, we didn't see uh things happen yet. So we are still like the
39:21 Speaker 00 the beaters and the early players in this market.
39:28 Craig S. Smith? How long does it take to train an avatar of a personal avatar for use on your platform?
39:36 Speaker 00 Yeah, so we provide all three types of avatar on our platform. So the first type is called instant avatar, where that will go to you. Upload a video instantly works. And for the second one is called a studio avatar. Upload a video, you will train on your video. And that takes about 12 hours, but everything happens automatically on the back end. And the third one is called ultra avatar. So with that avatar, you can control all the body motion, the hand movement, all the give much, give you much more control rather than just
40:22 Speaker 00 just the face and so on. You can control the body movement as well. With that one, it takes longer. So our our support team need to get involved to help you to create it. That might take like up to three days.
40:40 Craig S. Smith? Yeah. And and what's your vision for the future? I mean, you you know, the transformers have only been around since what, 2017? So not even 10 years. GPT-3 and onward have only been around for a few years and you guys have only been doing this, what, three years? When you look out five years, do you think the barrier between languages will will really disappear? I mean, what what do you envision?
41:19 Speaker 00 Yeah, that's that's a goal we are working on, right? So definitely, we want to remove the barrier of the language. Then so it's also very helpful for me personally. So we're definitely working towards it. And then we do believe that the barriers of language will be greatly reduced in next
41:45 Speaker 00 Three to five years, lots of the video translations will happen, lots of the real time translations will happen. So maybe in the next generation, they don't need to learn so many language anymore.
42:01 Craig S. Smith? That's right. That's right. Uh, and, and this Spidey researcher that I, I talked to seven years ago. Uh, he was imagining, uh, you know, an earpiece, uh, maybe connected by Bluetooth to your phone, where you could have a conversation, uh, and it would be translating, uh, the, the other person's speech live in your ear with a low enough latency that you could have a natural conversation. Uh, what, what's your thought on that?
42:41 Speaker 00 I think, uh, on the hardware side, that should be mature already. I think I already see some prototypes or some kind of that kind of product. How do the main thing is still on their AI side? How to ensure the translation accuracy? Uh, if you do voice call, I can't do that really well. And how to make it sound natural and how to make it to contain the emotion in the a translation, a reduce latency. So all of these are software related and AI related.
43:23 Speaker 00 On hardware side, I think that definitely do very do a open up, so.
43:27 Craig S. Smith? Yeah, and, and on, on emotion, for example, what are you guys working on right now to? Uh, what's your roadmap? Is emotion one of them?
43:39 Speaker 00 So emotion is a small piece in the current version, but we are working on improving the emotion in the translation and so on. So the emotion in the mainline use some of transfers from the original audio track to the new audio track with the emotions. But that's definitely a big piece that we've been working on. And in our first voice emotions on the avatars. And then next will be try to the video translation to copy emotion as well.
44:16 Speaker 02 Hola, meu nome é Eugeni e esta é uma demonstração da Equal Life Camera. O mais recente produto em nosso EcoLife Switch. Uma das primeiras soluções do mundo para geração de vídeo em tempo real. Isso inclui troca de rosto ao vivo, avatares ao vivo e tradução de vídeo em tempo real.
44:31 Speaker 02 Привет, меня зовут Евгений и это демонстрация Equal Life Camera. Последнего продукта в нашем EcoLife Suite. Одного из первых в мире решений для генерации видео в реальном времени. Это включает замену лица в реальном времени, живые аватары и перевод видео в реальном времени.
44:47 Craig S. Smith? Yeah, wow, uh, well, that's, that's all amazing. Uh, and I'm going to definitely try it if you can get me into the our cool studio. Was it that you need an invitation, which, which were you saying?
45:06 Speaker 00 Yeah, yeah, so cool night. Especially in the line camera, currently it need the invitation to get access to. And that's on the days and things are coming, so it's very exciting moment.
Previous #264 Amr Awadallah: Vectara's Mission To Make AI Hallucination-Free & Enterprise Ready Next #266 Andy Kurtzig: How Pearl Uses AI + Human Experts In Professional Services Page updated November 3, 2026 · transcript and summary are machine-generated
