CEO Stories with Ron Jaworski, Trinity Audio
Ron Jaworski’s Tel Avi-based company, Trinity Audio, is helping content providers not only convert “print” to audio using artificial intelligence, but create new revenue streams for content creators and larger, stickier audiences for that content.
Ron sees the case for audiofication of content as a stronger one than using video – where he started out. Audio is the optimal way for multitaskers to consume content. They can easily do it while driving, while performing mundane tasks or chores, and of course it offers accessibility for those who are blind, visually impaired, illiterate, or have low reading skills. Studies have also shown that an audio used is more likely to be a subscriber than a non-audio consumer and that they read as well as listen.
THE HUMAN TOUCH
The improvements in AI, making AI voices virtually impossible to distinguish from human ones, has made audio accessible in every sense, affordable – content can be audiofied in seconds, with the placement of a line of code in a file – and accounts for its growing acceptance. He cited a study where people were asked to listen to eight voices, and determine which were human and which were not. In 50 percent of the cases, people could not tell. Listeners split the other 50 percent evenly between human and AI voices. It turns out that all eight had been generated by AI.
Ron, who started out in video for the Internet, has always seen the web as having three pillars: the users, the publishers, and the advertisers. Audio just as easily as any other form of content – print, or video – lends itself to serving those three pillars. By offering something of value for all three you create a sustainable business.
Read Full Transcript
POC: Hi, this is Patricia O’Connell, welcome to CEO Stories, This is Capitalism. I’m talking this morning to Ron Jaworski, the CEO and co-founder of Trinity Audio. Trinity Audio is a company the specializes in using AI for getting written content into audio.
So, for all of us who are in the audio side of content, such as podcasters, we are interested in finding out what is going on with getting more audio content and Ron is here to tell us about that. So, Ron, thanks for joining us today on CEO Stories.
RJ: Thank you very much for having me.
POC: So, Ron, I’m curious about a few things. I started off my career in print journalism, remember print?
RJ: Yes, I heard about it.
POC: And you know then everyone had this fear of okay, print was going to go away because we didn’t need the physical thing anymore. But the point is the written word still exists as the written word, whether it is online or whether it’s on paper. But now, what you are doing with Trinity is taking those written words and turning them into audio words.
RJ: Yes, that’s right.
POC: So, what does a proliferation of audio content that is based on “the printed word” mean for lovers of printed content?
RJ: I think at the end of the day, the audio offering that Trinity Audio has, and audio in general, it is basically another way to consume content that you like, that you love. And it’s about mindset, first of all, and it’s about efficiency and it’s about multi-tasking.
And it is also about the way that you prefer to consume your content. Because people that like to read will keep on reading, although there is a video or an audio player on the website or on the app. People that like to watch videos will continue to watch videos because this is the way they like to consume their content. And there are a bunch of people that love audio and will consume either radio shows or podcasts or, of course, what we are also doing, which is creating audio experiences in AI based of course, as you said, on print content.
POC: So, talk to me about the AI component. Because one thing that people don’t like about AI, first of all, some people just don’t like the whole idea, artificial intelligence just scares them in general, but they often associate AI with robotic sounds, with non-natural sounding voices. So how are you getting past that with using AI to produce the audio content?
RJ: The idea of Trinity Audio started four years ago, it came to life actually somewhere around three years ago. And in the last I would say more than five, six, seven years, a lot of things that weren’t commoditized are now commoditized and especially we are talking about AI, NLPs and things like that.
If we remember, the basic building block for taking print and turning it into audio, which is text to speech solutions, they are with us since, I don’t know, 20 years ago. Okay, but the only problem is that most of them – as you said and said correctly – were robotic. Is sounded like a robot is talking or a machine is talking to us.
And what would happen with the commoditization of NLP processes and AI is that a text to speech engine today creates an amazing, amazing experience for people who listen, and it just improves more and more. So, if we made a major leap in the last three years ago and we continue exponentially to do leaps in the quality of the sound and the quality of the voice. This is on one hand.
On the other side, us as humans, we are communicating more and more with AI voices, whether it’s a smart speaker, voice assistance, voice bot, and whatnot and our ear becomes more and more tolerant to mechanical voices. So, we are really close to the point where AI voices would be as close to human as possible on one hand and our human ear will be, I would say, in sync with the consumption of AI voice.
And I want to add to that, and this is actually a test that we did more than two years ago – and voices are much more improved than two and a half years ago – we took eight different voices and gave it to a group of people and asked them to rank them either mechanical or human. And we got, from the eight voices we got 50 percent said we are not sure, either mechanical or either a human. We got 25 percent of them that said that’s definitely a machine. And the other 25, everybody said 100 percent, that is a human voice. The only thing that all of them didn’t know that all of the eight voices were mechanical voices, they were AI voices.
POC: So, we are reaching the uncanny valley when it comes to voices?
RJ: I believe so. I believe so, yes, definitely.
POC: Where they become human enough that it actually is almost a little confusing to people. A couple of questions that I’m just curious about, is there a preference toward male or female voices in AI?
RJ: That is a great question. I do see, we do see from a service that we are running that males tend to prefer women’s voices and in regard to women it’s almost a 50/50 split. So in regard to that we do see that it changes from country to country. It doesn’t stay the same. And we have to still try to find the correlation in regard to is it like a cultural thing or something like that. But it is definitely interesting.
POC: That also raises a question about accents. You have this very sophisticated, continental accent. It’s one of those great accents that if you heard it in a movie, you couldn’t quite place where you care. It’s like, hmm, is he from Europe, is he from the Mid-East, is he just one of those international spy types? And you’re located in Israel, correct?
RJ: Yes, in Tel Aviv, that’s right.
POC: So, what do you do about accents in different countries? Let’s say for the time being at least English is the majority language in the United States but we have all sorts of regional accents. Do people want the Boston Globe read to them in a regional Boston area accent? Do you read the New York Times to people in a New York accent? How do you figure that out?
RJ: So, localization is a great place to look at the development of AI and adaptive languages. So, we do support accents. Today we support 50 languages with over 400 different voices, whether it’s male, female and of course accents. So in regard to English, we don’t have localization for the U.S. but definitely we have U.S. accent versus U.K. accent versus Australian accent. And my guess is that in years to come we will also do a suite with regard to Boston or New York or Texas. It will be in that manner. It’s going there, it’s not there yet.
For example, with Spanish, we have a U.S. Spanish, or Mexican Spanish, or Argentinian Spanish. So, we are moving in that direction. I would say it’s more on the geo side of variations. And of course once we get deeper into that then we’ll go into states and maybe even cities. So it’s getting there. It’s in development like anything else.
POC: Okay, good to know. I have lived in two cities, New York and Boston, that have very, very strong accents so that’s why I am particularly curious how you’re going to tackle those. What is the business opportunity for content creators to spend the money on yet another way for people to consume content? Because obviously doing this is not free.
RJ: Yes. Not yet.
POC: So, there is a cost involved for people converting their content into audio. What is the business case for doing so?
RJ: So, there are several business cases. First of all there is – and this of course is the first thing that comes to mind – there is an accessibility aspect to it and you want to have the option for users that are either blind, visually impaired, illiterate, or low reading skills, people that for some sort of a reason are unable to read your content have the option to listen to it. This is the first thing.
There is a renaissance for audio, and we’ve seen it the last couple of years. And an interesting fact is that we see a boom in podcasting, but we also see an exponential growth in audio books. And you would have thought that one would come instead of the other.
In most cases, when we are in the mindset of consuming audio, we are probably doing something else, or our eyes are not free to watch something or read something. It’s going to be driving your car, it’s going to be cooking, jogging, you name it.
And I think that at the end of the day, for a content creator – and we see a lot of them start to talk about okay, what is my audio strategy – I hear a lot about audio and everybody is doing something new, should I do a podcast, should I do a library, what should I do because audio is becoming a thing, I want to be there for my audience.
POC: And should I do a podcast with video? You and I happen to be recording by video now but this video stays between us. We’re only going to be putting out the audio stream of it.
RJ: Exactly. Those are super valuable question. And should I also give the option for my users to watch the podcast that I’m recording? Maybe they want to watch it. I don’t know, maybe. It’s a good question and it should be asked.
It doesn’t really matter if you are a large publication or a small blog, if you are a big brand or a small blogger, it doesn’t really matter, we want to give you the easy option to be a gateway into creating a world of audio and an audio experience for the user. Because everybody is looking to consume audio in the relevant mindset, and you should give them this option. We just believe that the easiest way to do that is to start with Trinity Audio.
POC: So, when you say you’ve delivered the easiest way, talk to me about that. Is there something in the way that Trinity enables its clients to convert to audio? I’m not asking you to give away your proprietary secrets here, Ron but talk to me in a way that no one can steal them.
RJ: So, what I would say is that when we set forth to conquer the world wide web and have the option for anyone everywhere to listen to any kind of content that you want, we decided the most important thing is it needs to be super easy, no heavy lifting and scalable and robust.
And today our integration offers any kind of content creator to take our solution, embed it, and within five to ten minutes have all their content, as we call it, audiofied. And by audiofied it means that there will be a small play button or a full player with different features that any kind of user that will enter your webpage will understand the usage and will press play if they want to listen. And this is the basic building block of building the audio experience as we see it
Of course, we have much more robust product and complex product to deliver even a larger audio experience and distribute it and manage it and what not, but we do believe that the basic thing to develop is having the option to listen to an article. And this is where you start and from that the sky is the limit.
POC: Where does advertising come in here?
RJ: Great question. Let’s talk about monetization. So monetization is relevant for some, not for all. As we see it, the world wide web needs, or a website in general, needs to have a solution, an easy audio solution. So for a publication, for example, like for Newsweek for example, the McClatchy Group or whatnot, they have the option to monetize the traffic, the audio traffic, that you are creating for them, meaning basically press play on our player you will listen to an audio ad in the beginning, the end or the middle of the article. And this is where advertising comes in.
First of all, we have a joke over here – when God created the internet, He wanted to have a place to place ads within it. So we do understand that we need to create the option to monetize the traffic because the deal of the internet is that you get the content for free and we as the content creator get to advertise on top of it and we offer that solution within our product.
POC: Well, free-ish. Places do have paywalls; places do have subscriptions. So there’s free and there’s free.
RJ: That’s right, that’s right. That’s completely correct. But we do support all kinds of business model that is relevant for the content creator. We have a full tech stack that supports any kind of initiatives, whether it’s for monetization with ads or a subscription or whatnot, it’s for them to decide. We decided we will create the content, we will manage and distribute it for you, you just decide how you want to monetize it.
POC: Can they put ads in the middle of an audiofied story?
RJ: Yes of course, of course. For example, if we will take the blog of This is Capitalism and we will put our player over there and will decide that you want to embed ads in the middle of the blog we can do that. It will be super easy.
POC: Okay and again that’s an audio ad, obviously, because people are not going to stop listening, then go look at an ad and then go back to listening?
RJ: That’s right, that’s right. It’s an audio ad. And actually, if we talked earlier about the AI, we are also now experiencing AI audio read of advertising and we are seeing great traction. So not only the article itself is being AI read, the ad itself is being AI read and we are seeing great results with that as well.
POC: I believe one of the publications at MIT is doing a whole series about exploring AI innovations. And one of them I think is about audification or audiofication, I’m not quite sure which it is. But how was it done before AI?
RJ: So, narrators. That would be the easiest way, of course. It’s high quality, scalability is a bit of a problem in regard to that. But this is the way it has been done in the past.
POC: I’m just thinking for example something like the number of books in a public library. I have an e-subscription to the New York Public Library, in addition to a regular library card. Some books are available to me as audio books, some are not. I imagine a lot of that has to do with whether the book was actually published as an audio book originally.
But I’m imagining you could theoretically audiofy all of the books in the New York City Library, the books that were not originally published in that content, books that were published before audio as a technology even existed. How difficult would a project like that be? I have no idea how many books are in the New York City Public Library. Let’s, for the sake of argument, Ron, assume it’s one million books that have not been audiofied.
RJ: Yes. Yes. It’s something that can be done. I’m doing some calculation in my head in regard to the amount of pages one million books have and how much. But I think we can take all the New York Library in a couple of weeks and make everything audiofied. It’s something that will probably take, in the case that if you will use narrators, several years. And at the end of the day, I believe that all the content will be audiofied using AI.
POC: What are you finding is the compelling business reason that creators of content, creators of everyday consumable content that people are consuming information that makes them either smarter, better able to do their jobs, better informed for going out in the world?
What is the business case for the magazine publishers, the newspaper publishers, the journal publishers? When I say the journal, I mean professional journals not any specific newspaper. What is the business model and the business reason? Are they finding that they are getting new audience members? Audience is probably the best word since they are technically not readers.
RJ: So, I would say that in regard to journalism, to newspapers, magazines, things like that, they have a new channel to engage with their users, with their readers. And the fact that the user has an additional option to consume the content. A user that has the option to engage with audio is a more engaged user. He stays more time on the page, he engages with more web pages, his bounce rate is significantly lower.
We are talking like 100 percent. We check within our system, subscriber versus non-subscriber then we see that even the – let’s speak for journalism, let’s call it the heavy users, the heavy readers there is more likely for a user to be a subscriber if there is an audio option on the page and he will use it more frequently than a non-subscriber.
So audio is about stickiness, it’s about opening new hours throughs the day, a different client, audience or whatnot can consume the content. But, and this is important, we talk a lot about media and about publication, which the use case is obvious, yes, give me the option to listen when I don’t have time to read or when I can’t read or watch, but we do see as we move along that there are many, many use cases that we didn’t even think about that the clients are using.
For example, I can’t share the name of the client, but they are using our solution to actually create specific memos and basically distribute it within different groups that they have, like updates on things that are new within their community. And they do it on their web pages or they are doing it within a group where they use our solution to generate content and send it through WhatsApp or different messaging apps to a closed group.
We have users that took our solution and created eBooks. They took the eBooks and created an audio file. And we have the McClatchy folks that are using our solution to create flash briefings that are being updated on a daily basis from Spotify or Google Nest and Alexa in regard to smart speakers.
So the great thing about it is that people want to use it the way they think their audience would like to consume the content and they are doing crazy things and it’s great seeing that. We see a lot of innovation once we give the platform for our clients and they do crazy things with that and we are happy about it.
POC: This is very interesting to hear that there are other ways of clients using this. Because I am thinking about it in a very one-dimensional way, frankly. I’m thinking about it in terms of content creators who are providing an alternative to those of us who say gosh, I didn’t really have time to read everything I should have read today but I have two minutes to listen to that news story.
But you are also saying now that clients can do it, for example, if they just want to send a quick announcement out internally to people, or update them on something, they can grab their attention with their ears rather than their eyes? And it’s… I’m guessing there is some sort of urgency when people get an audio alert that they know that they have to listen to it?
RJ: So I would say that first of all, yes, you are right, the use cases for using audio are vast. You can use audio as you said just as a consumer, I don’t have time, let’s listen to that for two minutes. We do see a great engagement of readers with our solution, or I should say listeners, in long form content.
We are working with a website that has a vast amount of literature on their website and in many cases episodes of 10 minutes, 15 minutes, 20 minutes of AI read, and people are… and the completion rate on those audio files is amazing. So people are not only using that when they have to listen for something, they use it because it’s basically… In many cases, this is what we hear from the listeners – “it’s efficient, I can multitask.”
And a lot of them are just audiophiles and they love audio. They consume audio. By the way, they love also audio books and love podcasts. So it just makes sense to them to consume the content.
And another thing that we need to understand, that they are different groups of people, they are different clusters. And some of them prefer to use their eyes to consume content, some of them prefer to use their ears.
POC: All content does not lend itself to be turned into audio though I’m thinking – anything that’s rather heavy with graphs or charts. By the time you try to explain those graphs or charts, I might as well be looking at them.
RJ: That is 100 percent correct. Not any kind of content is audio relevant. And, as you said correctly, tables, graphs, in many cases things that are a bit more complex and not all of them can be easily turned into audio or easily consumed.
But in many cases… For example, we are now doing actually a test with education, a company, in regard to the way they want to serve their different courses online. And they are taking specific lessons in those syllabus and putting them online to give their students the option to consume it. There is some content that is relevant, some content that they can listen on the way to the college to the university and then we can go over the graphs and over the tables within the class itself.
So, it’s a mix and match of that and you need to understand the use case. Because different use cases mean different solutions.
POC: I think it’s safe to say that we will be seeing or rather hearing more audio content in the future, but audio is not going to be the only way content is consumed.
RJ: A hundred percent correct. And I think that one thing that we definitely need to mention and this is also I would say the other side of the coin of audio, which is voice, which is also a major part of what we call the audio revolution. And when we are talking about creating an audio world, an audio experience from content creators, this is the first step.
The voice layer of it is also super important because that would be the next stage. The conversation, it won’t be only a monologue where I am supplying you the option to listen to and to do other content, it will be also a discussion. You can ask questions, you can have different functionalities, you can a service or a test or to check out whether someone that listened to the audio file really got the main items or the main message that you wanted to deliver from audio.
And this is where we are going. I think it’s still in development, it’s something that we need to do but think about it when you want to deliver some sort of a message to your students, for example, and you can do and survey or test or examine the end of the specific content, asking them questions where they can answer with their voice to see whether they’ve got it or not. And if they didn’t get it, go over a specific paragraph or a specific.
And this is just a specific use case. There are 100 more that are relevant for basically the input and the output, the voice and the audio.
POC: So will there be actual interactivity, live interactivity?
RJ: Yes, yes, definitely. This is where we are heading.
POC: Okay so that’s really where the artificial intelligence part comes in?
RJ: And the creation of the audio itself is AI, the analysis of the textual part to understand what should be audiofied, the conversation of course is 100 percent AI, whether it’s the voice or the understanding of what is the question and what should be the answer.
POC: Well a brave new world. Ron, the last question I have to ask you is about the genesis of Trinity and frankly the name Trinity. What does Trinity signify?
RJ: So there is the honest answer and there is the funny answer. Where should I start?
POC: Oh, tell me both.
RJ: So I was a part of the start of it, we started in 2014, it was mainly into video, and it was sold to a public company here in Israel in 2016. And again, we started Trinity Audio under the umbrella of the company before we came out as Trinity Audio. And what we found out is, or what we thought, I would say, is that if a solution within the publishing landscape has value only if it serves the three pillars of the internet – the users, the publishers, and the advertiser.
And we call it the Holy Trinity of the internet, which is as I said the users, the advertisers, or the publisher. Once you have a product and you see it serves all those three pillars, then you have value within the ecosystem. You need to serve all the three. And we believe that Trinity Audio serves all those three pillars. The user gets something, the publisher gets something, and the advertiser gets something. This is the honest answer. The funny answer is of course it’s always funny to have three Jews in Israel talking about the Holy Trinity.
POC: Well, that’s a little bit of what I was getting at there, Ron. Notice I asked about “genesis” there…
RJ: Yeah, right. [Laughs.]
POC: So, I set you up there a little bit. Well, Ron, I hope that the next time we speak it will actually be live and I will not be speaking with an AI bot that you set up for me.
POC: Okay, Ron, thanks so much for joining us today to talk about Trinity Audio on This is Capitalism.
RJ: Thank you very much. It was my pleasure. And I hope to see you face to face soon.
POC: Likewise. Thanks, Ron. And you can find out more about Trinity Audio at www.trinityaudio.ai. That’s trinityaudio.ai. I’m Patricia O’Connell for This Is Capitalism, CEO Stories.
About the Series: Featured stories from the intersection of the free market and entrepreneurial success. Here we speak with leading CEOs, academics, philanthropists and up and comers on their contributions and perspectives on the American economy.
About Patricia O’Connell: Patricia O’Connell serves as Editor in Chief of “This Is Capitalism,” a content site sponsored by Stephens Inc., and is host of the site’s podcast series, “CEO Stories.” Patricia, a former editor at BusinessWeek and a New York Times best-selling author, brings her experience as a journalist and her passion for storytelling to “This Is Capitalism.”