#272 Spatial Audio, IMDb Honors Dubs, Kindle AI Translations, Startup Rounds

0:00Florian:Spatial audio can double comprehension in live AI translation by placing the translated voice where the speaker appears on screen.

0:12Hey everyone.

0:13Florian:It's good to be back and catch you up on a few weeks worth of language and solutions developments. Hey there, Esther. Hey, Florian. It's been a while since we had our last news catch up. I checked it was late October, 2025. Yes. It's been a while. It's high time. Time to come back. Absolutely. Well, we're still very much amped up from SlaterCon remote yesterday. Huge turnout, over 500 registrations. Thanks to all the speakers, panelists, and participants. But now we're going to go back to in-person again. So great news. SlaterCon London, happening in May 2026, is now open for registration. and we are offering a very, very attractive super early bird discount.

1:00Florian:So you want to go there now, not in April. And please do that because it helps with our planning. And anyway, I mean, joining this is worth literally just for the food. Remember the food? Yeah, the food is remarkable. The food at the Nobu. I mean, it's the best buffet in the world. Chef Nobu. Last time he was there, Chef Nobu. Yeah, he hosted some type of session. Yeah, fine, three course dining or something. I was a bit starstruck. Yeah, we're not going to offer you any food made by Slater, but we're going to offer you three additional tracks in addition to, I don't know, it's going to be about a dozen tracks that we've had last year, but this time we're going to add another three.

1:41Florian:One is going to be startups, language, AI research, and NGO happening in a separate room in addition to kind of the main stage. So it's going to be great. I think we're going to even increase the total participant count. We're going to max out the locale there to about 250 people, 30 to 40 speakers and panelists. We're also getting a ton of interest from potential partners. So if you do plan to partner, do get in touch now because interest is very, very strong. And I think we're going to be allocating these slots very, very soon. I don't think we prepared an agenda to go through because we have so many things to go through.

2:20Florian:so I don't want to waste time telling you what we're going to be talking about, but there's a ton of stuff obviously that happened in the past five to six weeks that we want to catch you up on. First, good news. What's the good news, Esther? Some good news for once. Yeah, I mean, we were going to discuss a little bit about IMDB and the fact that it is now recognizing dubbing artists in a new credit category. So that's quite exciting. I had to look this up. I know this is something I didn't really know. So IMDB was founded in 1990 and apparently began as a crowdsourced list of movie facts and cast information.

2:59Florian:And then I did not notice until we published our article that Amazon bought it in 1998. I didn't know. 1998. Yeah, I was not aware. That's like four years after Amazon was founded. And then it grew into the world's largest film and TV database. I don't know why Amazon thought it was a good idea to buy this in 1998. I thought they were very kind of disciplined on e-commerce back in the day. Maybe as, I'm sure there's something like distributing films instead of all the books at that time. I mean, you know, the younger listeners may not remember that Amazon was an online bookstore. I wasn't totally up on sort of like corporate incidents back in 1998 having just started secondary school.

3:43Florian:Even I was too young to care about corporate stuff in 1998. Yeah, yeah. All right. So basically on its mission or its mission is to document every part of the global screen entertainment, ensure transparent, accurate credit for everyone involved. And that's why they have added now 12 new professional credit categories and roles that were previously lumped into additional crew now get proper recognition. So things like choreography, color, craft services, health and safety, intimacy coordination. That sounds interesting. Legal production, production, finance, accounting, property, publicity, puppetry, and voice actor dubbing is now a category.

4:25Florian:So now they recognize 45 total professional credit categories. And yeah, that's good news. So initially, IMDb will now accept dubbing credits in English, French, German, Hindi, Italian, Portuguese, and Spanish. And this gives multilingual voice actors and dubbing studios greater industry visibility, which is good. I mean, I think even every time you... Along with the puppets. Along with the puppets and the intimacy coordination. I'm going to look that up. I think, you know, if you do watch in Netflix all the way through the credits, they do list a lot of the dubbing studios and some of the individual voice actors, so it's good that you have this now kind of cross industry.

5:05Florian:Cool. Next, we want to catch you up on is that Coursera has a new CEO, new-ish CEO, And that individual named Greg Hart bets big on AI translation and AI dubbing for driving growth because Coursera hasn't been doing so good. What's Coursera for people who might not know? What is Coursera? How would you describe it? It's like an online learning company. So they have tons of sort of e-learning courses. Courses. Name-like-like. Okay. On everything, right? But the shares, the company has lost like 80 % of their market cap since the 2021 IPO. Good time to IPO. 2021, COVID, online, remote, big hype, everything's online.

5:56Florian:Nobody does anything in person anymore. You know, go IPO with a company that offers e-learning. But unfortunately, like so many other COVID plays in the stock market, Coursera also lost a ton of money. Not money, sorry, market cap. And now the new CEO is trying to revive growth, double down on multilingual content, because that's a quick one, right? You produce in English, you dub it. And let me read the quote here of the newish Cursor CEO. He says, the cost of translating and dubbing is unbelievably cheaper because of AI. And also the outcomes are better. Not surprisingly, a learner learns better in their native language than they do in English.

6:37Florian:And so we see that when we translate content into more and more languages, we get better engagement and better completion rates from learners in those geography and those languages. A couple of things. Not surprisingly, a learner learns better in the native language than they do in English. Well, what if they don't speak English at all? I love the English centricity of this all. Like, okay, there are many billions of people that do not speak English, so they cannot learn better. They could not learn anything if it's not in their language. Isn't he just pointing out that the content is currently all in English and they're like a US company, I'm assuming.

7:15Florian:I mean, the actual quote is a learner learns better in their native language than they do in English. Better. Well, if you don't speak English, you learn nothing. So I'm just being, I don't know, snarky. Yeah. Okay. So, well, this is, I guess, good for anybody who's doing the AI dubbing over at Coursera because they went from 100 to 600 plus courses in five languages and they target a thousand plus by the year end. They've been using AI translation since 2023 and now there's over 5 ,600 courses that they offer there on the translated side, not the dubbed side. Now, did you know, and there's a bit of full circling going on here because The actual co-founder of Coursera is Andrew Ng, and Andrew Ng is literally the godfather of part of the current AI boom.

8:11Florian:I mean, he was an AI researcher, co-founder of Google Brain, absolute, basically legend in AI. On Twitter as well, right? Is he on Twitter? Yeah. I mean, but he kind of stayed a researcher. I don't think he probably doesn't seem to like the corporate thing too much, like corporate, like just the big company corporate thing. I mean, he co-founded Coursera and now I'm not sure if he's still active and he's not like leading any major divisions at Google or OpenAI or whatever other AI company. Anyway, that's just a quick aside. Andrew Ng co-founded it and now they're using some of the results of his early research to enter new markets.

9:01Florian:Now, would you want to read AI translated fiction, Esther? I think not. Not? What if you didn't know that it was AI translated and you just read it? Then I would not care. I mean, I feel like the decision would have been taken away from me. That is true. Yeah, I don't know. Not particularly, not as advertised. I mean, I wouldn't shy away from it necessarily, but maybe I'd need to have some good reviews first. Did you have a Kindle? Do you own a Kindle? I don't. I have Kindle, like, what do you call it? The Kindle app on iPads or something. Yeah, I mean, because Amazon now says that they're launching AI translated eBooks on Kindle and that they promise a turnaround of 72 hours to produce your AI translated Kindle ebook promised to the authors, that is.

9:58Florian:Yeah, I don't want to go too much into detail. It's literally just a question like, okay, so you're an author. You spend, I don't know, half a year, a year, or at least a few months putting together a book, maybe even longer, and then you just click a button and let Amazon AI translate your work. I think especially fiction, I don't know, fiction, rather than something sort of factual. Like maybe they think it's less dangerous to let AI translate fiction rather than some sort of, you know, biographical. Plus they get the author's consent, I would assume, right? So it is a quick win for Amazon. Like, okay, you have to tech, you have to consent.

10:40Florian:No one's going to die if there is an error. Yeah. So why not? And then now it'll be super interesting if this is actually resonating with readers. So I really think we should check back in a year from now if this actually works. Yeah, maybe we should actually force ourselves to read AI translated fiction as a sort of research task. Yes. We let the AI summarize this, give us an action item and drop it into our to-do list. Okay. Before we go to a bunch of stories around funding, I think it's mostly funding. I want to talk to you and I want to open up a quick research corner, Language AI research corner, because we came across this really interesting study led by Margarita Giletta, who is a UC Berkeley researcher, but then did an internship at Microsoft and published kindly the results in a YouTube video.

11:39Florian:And so they examined how spatial audio affects comprehension in AI live speech translation. And so they used a controlled multilingual dialect tests with 47 participants. Spatial audio. I don't know what that means. Yeah. Well, you hear where the sound's coming from, right? I mean, if you're in a room, you have spatial audio. I don't know where you're okay. I mean, like generally you have spatial audio. Right. So spatial audio can double comprehension in live AI translation by placing the translated voice where the speaker appears on screen. You know, I think with the new headphones you can do that.

12:24Florian:Massively improves tracking who said what, especially during rapid back and forth dialogue. One ear translation performs worst. Listeners find it confusing, tiring, less accurate. And then, you know, voice timbre matters. Mixed gender speaking speakers are easier to follow. Yeah. So practical takeaways for the platforms are aligned translated audio with speaker tiles, give users control over original audio levels and maintain clear distinct speaker voices, but then there's the whole challenges like the latency, overlapping speech, messy real world meeting conditions, you know, that still limit live AI translation performance, which is what a bunch of startups and LSIs and LTPs are working on, I mean, that whole messiness of life is just super difficult.

13:12Yeah.

13:13Florian:Now, I guess we have a couple of startups that are working on this, right? That are raising some money. Yeah. We have a couple of funding stories and one former startup acquired by Giant. Giant. In terms of a roundup on the funding news from past month or so. So we have news that a company called No Barrier has raised$2 million. No Barrier was actually the CEO and co-founder of AR was actually on the panel yesterday at SlaterCon Remote. So that was great to hear all about what they're working on. But just as a quick summary, No Barrier is a healthcare AI tech startup. So it's focusing on AI live speech translation and AI interpreting.

14:02they announced that they raised to 2.7 million in a seed round I think that was in October that the round actually closed and that brings their total funds raised to just over 3 million then we had news I think we sort of shared this news a few weeks ago now that Cisco acquired a dubbing startup called Easy Dubs um it's interesting sort of background to the company i think it was a white y combinator alum um raised around 4 million in 2024 based in san francisco and so he's dubs ai startup that's focusing on real-time speech translation um so i guess not dubbing per se as the name might imply, but a real-time speech, meaning that it's probably most suited to the WebEx suite of technology.

15:06So as we've seen sort of other large companies that have this kind of conferencing technology.

15:13Florian:WebEx, you're right. Rolling out, yeah. I forgot about WebEx. When was the last time I used WebEx? Yeah, so it's like the product launch into WebEx that will be the kind of live speech translation equivalent to what we've seen the others Google, Zoom, etc. Okay. So that seems to be the rationale behind the acquisition. Cisco that acquired EZdubs said that essentially it's filling a gap, a critical gap, and says that it is marking the fact that the company is all in with AI in collaboration. Those are big words. All in and critical gap. All in, okay. You're basically acquihiring a tiny company and then you're saying it marks your all-in moment for AI in collaboration.

16:06Florian:All right, Cisco. Just to mention, there was the Women in Localization founder was part of the E-Dubs board. Yeah. So she was the product advisor to the team there. And then final sort of funding story that we're going to touch on today, AudioShake raising$14 million. This is another audio AI startup. So what we had, we've had, yeah, healthcare interpreting live speech AI with no barrier. Then we had this real-time speech translation solution with EasyDubs. Now we're talking about audio AI startup. up and so very much focusing on um yeah voice interpreting speech generally um within the environment so audio ai um the the startup called audio shake raised 14 million dollars in series a um this one is quite interesting i mean i was trying to wrap my head around what it was or what it does exactly but they said they use proprietary ai models to separate recordings into distinct tracks so imagine like film or music or some kind of broadcast scenario you're splitting up the recordings into the vocal component like the instrumental component and and the effects um so it might not sound super related to uh you know the the kind of services or solutions that we care about per se but actually they say that by isolating clean dialogue and by removing background noise, it actually supplies better inputs for things like AI audio transcription, live captions, live translated captions, AI dubbing, AI subtitles, and of course, AI live speech translation.

17:54Florian:Yeah, I mean, let me bring it back to what we just said before with Margarita's Microsoft project, where basically the big challenges that remain, there was latency, overlapping speech, and messy real world meeting conditions. So, this takes care of, I guess, overlapping speech and messy real world meeting conditions to some degree. So, a very interesting company and they're, you know, 14 million, we'll get them somewhere. Yeah, we should get them on the podcast. That was interesting. I mean, it was, you know, similar-ish to Iola that we had on the podcast last week, so there seems to be a big rush into the just, I mean, generally doing things with speech, so very interesting.

18:39Florian:All right, now let's talk about one public company or kind of public-ish company we never spoke about called OneMeta. Now they are listed, but they're not listed on a major exchange like Nasdaq or New York Stock Exchange, but they're listed on something called like basically an over-the-counter desk on the OTC Markets Group. It's basically OT markets group. It's basically where smaller or early stage companies trade when they're not on big exchanges. So, you know, think of it as kind of a more flexible, easier entry marketplace with a wider range of company qualities. Investors have to do a bit more research and research you can do if you go and look at their filing OneMeta.

19:23Florian:So OneMeta is like an early stage AI language LTP. I think they're pretty heavy in interpreting like OPI and things like that. So basically, yeah, US interpreting. Now, they are growing revenues but they also have a ton of costs. So we're getting it from their results, right? So apparently in Q3 2025, they hit almost half a million dollars in revenue compared to like basically three and a half thousand in Q3 last year. So basically went from having no revenue to having half a million revenue. And year to date in the nine months, they have now 1.3 million in revenue. They have solid growth profit, but they also have a ton of operating expenses as you'd expect in a startup.

20:20Florian:up and they have very little marketing, only 7K for the quarter. That's probably why we haven't really had them on our radar because they have done very little marketing. The net loss though in Q3 was almost over$800 ,000, a little better than 900K last year, but still pretty bad. Year-to-date net loss,$2.5 million down from$2.9 million. So basically that TLDR is a company continues to burn cash faster than it generates despite revenue traction, yeah, it needs to get some money from somewhere, I guess, at some point, so we'll be tracking it. This was just a quick kind of summary of their financials, but like it's just something we've never really looked at here on Slaterpot, that company and their financials and they do have the questionable honor of having to publish quarterly results now, now that they're public.

21:19Florian:So yeah, we'll be tracking it and wish them good luck in, you know, narrowing the losses and getting the revenues up very quickly. Otherwise, I think it's going to get a little difficult. Now, let's talk about the customer side. So one big customer of language solutions integrators and language technology platforms is the UK Health System, the National Health Service. And they put out a giant tender again. Or no, the tender came into effect, that framework contract came into effect. Yes, they've effectively announced now who the contract awards go to. I think we originally covered the tender itself back in January 2025.

22:04So this is the results effectively of the RFP. So, yeah, pretty sizable. We have contact value of 73 million pounds, which is roughly 90 million dollars over two years. Also an option to extend for a further two years to 2029. That would take us. This is a an agreement for language services, as you said, with the NHS, the National Health Service, specifically the Shared Business Services Office. so they tend to do a lot of the procurement for the NHS. The framework came into effect I think just at the start of November and it's valid for the NHS as well as other public agencies. It covers services like in-person, telephone and video interpreting as well as document translation they call it which I think that bracket also potentially includes web and digital translation as because they've done a big sort of digital push as well, the NHS, just as a side point.

23:13Like now every GP practice or surgery is supposed to allow sort of online booking. So you can imagine, you know. Does it work? I haven't actually tried. I mean, I think it's only mandatory since October. So mercifully, I don't think I've had to use the NHS service since then.

23:30Florian:That's good. And then, yeah, one final sort of lot, which is the word they use to kind of describe the batch of services as a one-stop shop. So this is sort of a managed service where the users could come and buy any one of those services or all of those services under the one bracket. So, yeah, we had, I think, more than 30 companies were awarded onto the framework contract. Some well-known names like Aclab, also Language Line Solutions, And I think we, I don't know how this was pointed out exactly, whether it sort of forms part of the framework contract information documents. But we pointed out in our article that it's related to the Equality Act of 2010, as well as the NHS Act of 2006.

24:20So there is kind of like mandated, you know, call it regulations within the UK that are mandating equitable care and informed consent that are kind of, I suppose, motivating and driving the use and the procurement of these services. So talk about sort of regulatory drivers. This would be a strong example of that.

24:40Florian:All right. Yeah, you mentioned LanguageLine, which we also heard from indirectly because Teleperformance, their parent company published Q3 results. Very steady-ish in terms of just a little bit of revenue growth. But what we really care about is their Teleperformance's specialized services division, which includes interpreting. That one grew by 2.6%. They did say that the US market remained very volatile. as you'd expect given all the regulatory and English as a national language kind of stuff which may impact interpreting. And yeah, if you look at the share price, huge pressure on teleperformance generally.

25:28Florian:I mean, it looks like a mountain goes up. Basically, they're back where they were in 2020, sorry, 2015, so it's like a 10 year round trip now because hundreds of thousands of people, contact centers, et cetera, all the things that like on the face of it, AI is pressuring. We also heard from AMN Healthcare, which is the owner of, they have a language services division, which same, generated flat revenues like down 1%,$75 million in Q3, flat year-on-year, and as I said, down at 1%, so basically stable. But they expect lower margins in Q3. And let me just pull this up here. They said there were seasonal declines in language services minutes and lingering client transitions will weigh on Q4 results.

26:28Florian:And they see some pricing compression and tougher immigration trends. Yeah, no kidding. They do say that they remain optimistic though, as they should when they talk to investors. All right. That was, I don't know, what was it? A dozen individual items, news stories, 15, 20, I don't know, but we wanted to catch you up and give you something to chew on. If we, I don't know, maybe we get a chance to record another podcast before the year end will be quite challenging though. So another guest episode coming up though. Very, very, very cool one recording this week. All right. So let me leave you with two calls to action.

27:12Florian:Now first, go and register for SlatorCon London. And if you do plan on partnering with us at SlatorCon London, you do really want to get in touch now. We're distributing some of these highly coveted partnerships right now. All right. Thanks, Esther. Thank you.