See all Lex Fridman transcripts on Youtube
Dan Kokotov: Speech Recognition with AI and Humans | Lex Fridman Podcast #151
1 hours 28 minutes 46 seconds
🇬🇧 English
Speaker 1
00:00
The following is a conversation with Dan Kokorov, VP of engineering at Rev.ai, which is by many metrics, the best speech to text AI engine in the world. Rev in general is a company that does captioning and transcription of audio by humans and by AI. I've been using their services for a couple of years now and planning to use Rev to add both captions and transcripts to some of the previous and future episodes of this podcast to make it easier for people to read through the conversation or reference various parts of the episode, since that's something that quite a few people requested. I'll probably do a separate video on that with links on the podcast website so people can provide suggestions and improvements there.
Speaker 1
00:48
Quick mention of our sponsors. Athletic Greens, all in 1 nutrition drink. Blinkist app that summarizes books. Business Wars podcast and Cash App.
Speaker 1
01:00
So the choice is health, wisdom or money. Choose wisely my friends. And if you wish, click the sponsor links below to get a discount and to support this podcast. As a side note, let me say that I reached out to Dan and the Rev team for a conversation because I've been using and genuinely loving their service and really curious about how it works.
Speaker 1
01:22
I previously talked to the head of Adobe Research for the same reason. For me, there's a bunch of products, usually software, that comes along and just makes my life way easier. Examples are Adobe Premiere for video editing, iZotope RX for cleaning up audio, AutoHotKey on Windows for automating keyboard and mouse tasks, Emacs as an IDE for everything, including the universe itself. I can keep on going, but you get the idea.
Speaker 1
01:50
I just like talking to people who create things I'm a big fan of. That said, after doing this conversation, the folks at Rev.ai offered to sponsor this podcast in the coming months. This conversation is not sponsored by the guest. It probably goes without saying, but I should say it anyway, that you cannot buy your way onto this podcast.
Speaker 1
02:13
I don't know why you would want to. I wanted to bring this up to make a specific point that no sponsor will ever influence what I do on this podcast, or to the best of my ability, influence what I think. I wasn't really thinking about this, for example, when I interviewed Jack Dorsey, who is the CEO of Square, that happens to be sponsoring this podcast, but I should really make it explicit, I will never take money for bringing a guest on. Every guest on this podcast is someone I genuinely am curious to talk to or just genuinely love something they've created.
Speaker 1
02:48
As I sometimes get criticized for, I'm just a fan of people, and that's who I talk to. As I also talk about way too much, money is really never a consideration. In general, no amount of money can buy my integrity. That's true for this podcast, and that's true for anything else I do.
Speaker 1
03:07
If you enjoy this thing, subscribe on YouTube, review it on the Apple podcast, follow on Spotify, support on Patreon, or connect with me on Twitter, Alex Friedman. And now, here's my conversation with Dan Kokorov.
Speaker 2
03:23
You mentioned science fiction on the phone, so let's go with the ridiculous first. What's the greatest sci-fi novel of all time in your view? And maybe what ideas do you find philosophically fascinating about it?
Speaker 3
03:37
The greatest sci-fi novel of all time is Dune, and the second greatest is the Children of Dune, and the third greatest is the god emperor of doing so. I'm a huge fan of the whole series. I mean, it's just an incredible world that he created.
Speaker 3
03:53
And I don't know if you've read the book or not.
Speaker 2
03:55
No, I have not. It's 1 of my biggest regrets, especially because the new movie is coming out. Everyone's super excited about it.
Speaker 2
04:03
It's ridiculous to say, and sorry to interrupt, is that I used to play the video game, used to be Dune. This, I guess you would call that real-time strategy.
Speaker 3
04:14
Right, right, I think I remember that game.
Speaker 2
04:15
Yeah, it was kind of awesome, 90s or something. I think I played it actually when I was in Russia.
Speaker 3
04:20
I definitely remember it. I was not in Russia anymore. I think at the time that I used to live in Russia, I think video games were about like the specification of Pong, I think Pong was pretty much like the greatest game I ever got to play in Russia, which was still a privilege, right, in that age.
Speaker 2
04:35
So you didn't get color? You didn't get like a...
Speaker 3
04:38
Well, so I left Russia in 1991,
Speaker 2
04:40
right? 1991, okay. So
Speaker 3
04:42
I always wanted to feel like a kid because my mom was a programmer. So I would go to her work, right? I would take the Metro.
Speaker 3
04:49
I got her work and play like on, I guess, the equivalent of like a 286 PC, you know?
Speaker 2
04:54
Nice, with floppy disks. Yes. So OK, but back
Speaker 3
04:57
to Dune. What do you get? Back to Dune.
Speaker 3
05:00
And by the way, the new movie I'm pretty interested in, but the original... You're skeptical. I'm a little skeptical. I'm a little skeptical.
Speaker 3
05:07
I saw the trailer. I don't know. So there's a David Lynch movie, Dune, as you may know. And I'm a huge David Lynch fan, by the way.
Speaker 3
05:14
So the movie is somewhat controversial, but it's a little confusing, but it captures kind of the mood of the book better than I would say, like, most any adaptation. And like, Dune is so much about kind of mood and the world, right? But back to the philosophical point. So in the fourth book, God Emperor of Dune, there's a sort of setting where Leto, 1 of the characters, he's become this weird sort of God Emperor.
Speaker 3
05:41
He's turned into a gigantic worm. I mean, you kind of have to read the book
Speaker 2
05:44
to understand what that means. So the worms are involved.
Speaker 3
05:46
Worms are involved. You probably saw the worms in the trailer, right?
Speaker 2
05:49
And in the video game.
Speaker 3
05:50
So he kind of like merges with this worm and becomes this tyrant of the world and he like oppresses the people for a long time, right? But he has a purpose. And the purpose is to kind of break through kind of a stagnation period in civilization, right?
Speaker 3
06:05
But people have gotten too comfortable, right? And so you kind of oppresses them so that they explode and like go on to colonize new worlds and kind of renew the forward momentum of humanity, right? And so to me, that's kind of like fascinating, right? You need a little bit of pressure and suffering, right?
Speaker 3
06:22
To kind of like make progress, not get too comfortable. I don't know, maybe That's a bit of a cruel philosophy to take away, but.
Speaker 2
06:33
That seems to be the case, unfortunately. Obviously, I'm a huge fan of suffering. So, 1 of the reasons we're talking today is that a bunch of people requested that I do transcripts for this podcast and do captioning.
Speaker 2
06:52
I used to make all kinds of YouTube videos and I would go on Upwork, I think,
Speaker 3
06:58
and
Speaker 2
06:58
I would hire folks to do transcription and it was always a pain in the ass, if I'm being honest. And then I don't know how I discovered Rev, but when I did, it was this feeling of like, holy shit, somebody figured out how to do it just really easily. I'm such a fan of just when people take a problem and they just make it easy.
Speaker 2
07:31
There's so many things in life that you might not even be aware of that are painful. Then Rev, you just like, give the audio, give the video, you can actually give a YouTube link. And then it comes back like a day later or 2 days later, whatever the hell it is with the captions, you know, all in a standardized format. That was, I don't know, it was truly a joy.
Speaker 2
08:00
So I thought I had, you know, just for the hell of it, talk to you. 1 other product, it just made my soul feel good. 1 other product I've used like that is for people who might be familiar is called iZotope RX. It's for audio editing.
Speaker 2
08:19
That's another 1 where it was like, you just drop it. I dropped into the audio and it just cleans everything up really nicely. All the stupid, like the mouth sounds and sometimes there's a background like sounds due to the malfunction of the equipment. It can clean that stuff up.
Speaker 2
08:41
It has a general voice denoising. It has like automation capabilities where you can do batch processing and you can put a bunch of effects. I mean, it just, I don't know, everything else sucked for like voice-based cleanup that I've ever used. I've used Audition, Adobe Audition, I've used all kinds of other things with plugins.
Speaker 2
09:02
You have to kind of figure it all out. You have to do it manually. Here, it just worked. So that's another 1 in this whole pipeline that just brought joy to my heart.
Speaker 2
09:12
Anyway, all that to say is, Rev put a smile to my face. So can you maybe take a step back and say, what is Rev and how does it work? And Rev or Rev.com? Rev, Rev.com.
Speaker 3
09:28
Same thing, I guess. We do have Rev.ai now as well, which we can talk about later.
Speaker 2
09:34
Do you have the actual domain or is it just a...
Speaker 3
09:37
The actual domain, but we also use it kind of as a sub-brand. So we use Rev.ai to denote our ASR services, right? And Rev.com is kind of our more human and to the end user services.
Speaker 2
09:50
So it's like WordPress.com and WordPress.org, they actually have separate brands that like, I don't know if you're familiar with what those are. Yeah, yeah, yeah. They provide almost like a separate branch of-
Speaker 3
10:00
A little bit, I think with that, it's like WordPress.org is kind of their open source, right? And WordPress.com is sort of their hosted commercial offering.
Speaker 2
10:08
Yes.
Speaker 3
10:09
And with us, the differentiating is a little bit different, but maybe a similar idea.
Speaker 2
10:12
Yeah. Okay. So what is Rev?
Speaker 3
10:15
Before I launch into what is Rev, I was gonna say, you know, like you were talking about Rev was music to your ears, your spiel was music to my ears, to us, the founders of Rev, because Rev was kind of founded to improve on the model of Upwork. That was kind of the original, or part of their original impetus. Like our CEO, Jason, was an early employee of Upwork.
Speaker 3
10:39
So he's very familiar with their Upwork company. And so he was very familiar with that model and he wanted to make the whole experience better because he knew like when you go at that time, Upwork was primarily programmers. So the main thing they offered us, if you want to hire, you know, someone to help you code a little site, right. You could go on Upwork, You could like browse through a list of freelancers, pick a programmer, have a contract with them and have them do some work.
Speaker 3
11:07
But it was kind of a difficult experience because for you, you would kind of have to browse through all these people, right? And you have to decide, okay, well, is this guy good? Or is somebody else better? And naturally, you're going to Upwork because you're not an expert, right?
Speaker 3
11:24
If you're an expert, you probably wouldn't be getting a programmer from Upwork. So how can you really tell? So there's a lot of potential regret, right? What if I choose a bad person?
Speaker 3
11:34
They're gonna be late on the work. It's gonna be a painful experience. And for the freelancer, it was also painful because half the time they spent not on actually doing the work, but kind of figuring out how can I make my profile most attractive to the buyer, right? They're not an expert on that either.
Speaker 3
11:51
So like, our idea was, let's remove the barrier, right? Like, let's make it simple. We'll pick a few verticals that are fairly standardizable. Now, We actually started with translation, and then we added audio transcription a bit later.
Speaker 3
12:05
And we'll just make it a website. You go, give us your files, we'll give you back the results as soon as possible. Originally, maybe it was 48 hours, then we made it shorter and shorter and shorter.
Speaker 2
12:18
Yeah, there's a rush processing too.
Speaker 3
12:19
There's a rush processing now. And we'll hide all the details from you, right? Yeah.
Speaker 3
12:26
And like, that's kind of exactly what you're experiencing, right? You don't need to worry about the details of how the sausage is made.
Speaker 2
12:31
That's really cool. So you picked like a vertical, by vertical you mean basically a service.
Speaker 3
12:37
A service category.
Speaker 2
12:39
Why translation? Is Rev thinking of potentially going into other verticals in the future? Or is this like the focus now is translation, transcription, like language?
Speaker 3
12:50
The focus now is language or speech services generally, speech to text, language services. You can kind of group them however you want. But we originally, the categorization was work from home.
Speaker 3
13:05
So work that was done by people on a computer, you know, we weren't trying to get into, you know, task rabbit type of things. And something that could be relatively standard, not a lot of options. So we could kind of present the simplified interface, right? So programming wasn't a good fit, because each programming project is kind of unique.
Speaker 3
13:24
We're looking for something that transcription is, you have 5 hours of audio, it's 5 hours of audio. Translation is somewhat similar in that you can have a 5 page document, and then you just can price it by that. And then you pick the language you want. And that's mostly all that is to it.
Speaker 3
13:42
So those were a few criteria. We started with translation because we saw the need. And we picked up kind of a specialty of translation where we would translate things like birth certificates, immigration documents, things like that. And so they were fairly, even more well-defined and easy to kind of tell if we did a good job.
Speaker 2
14:08
So you can literally charge per type of document? Was that the, so what is it now? Is it per word or something like that?
Speaker 2
14:15
Like how do you measure the effort involved in a particular thing?
Speaker 3
14:21
So now, like for audio transcription, right, it's per audio unit. Well,
Speaker 2
14:25
that, yes.
Speaker 3
14:26
For our translation, we don't really actually focus on that anymore, But back when it was still a main business of Revit, it was per page, right, or per word, depending on the kind of...
Speaker 2
14:36
Because you can also do translation now on the audio, right?
Speaker 3
14:40
Mm-hmm, like subtitles. So it would be both transcription and translation. That's right.
Speaker 2
14:45
I wanted to test the system to see how good it is. To see like how, well, is Russian supported?
Speaker 3
14:52
I think so, yeah.
Speaker 2
14:54
It'd be interesting to try it out. I mean, 1 of the- But now it's only
Speaker 3
14:57
in like the 1 direction, right? So you start with English and then you can have subtitles in Russian. In Russian.
Speaker 3
15:01
Not really the other way.
Speaker 2
15:02
Got it, because I'm deeply curious about this. When COVID opens up a little bit, when the economy, when the world opens up a little bit.
Speaker 3
15:10
You wanna build your brand in Russia?
Speaker 2
15:12
No, I don't. First of all, I'm allergic to the word brand. It's terrible.
Speaker 2
15:17
I'm definitely not building any brands in Russia. But I'm going to Paris to talk to the translators of Dostoevsky and Tolstoy. There's this famous couple that does translation. And I'm more and more thinking of how is it possible to have a conversation with a Russian speaker because I have just some number of famous Russian speakers that I'm interested in talking to.
Speaker 2
15:45
And my Russian is not strong enough to be witty and funny. I'm already an idiot in English. I'm an extra level of like awkward idiot in Russian, but I can understand it, right? And I also like wonder how can I create a compelling English Russian experience for an English speaker?
Speaker 2
16:06
Like if I there's a guy named Gregoriy Perlman, who's a mathematician, who obviously doesn't speak any English. So I would probably incorporate like, a Russian translator into the picture. And then it would be like a not to use a weird term, but like a 3, like A33 person thing where it's like a dance of like, I understand it 1 way, they don't understand the other way, but I'll be asking questions in English. I don't know.
Speaker 2
16:38
I don't know the
Speaker 3
16:39
right way. It's complicated.
Speaker 2
16:40
It's complicated, but I feel like it's worth the effort for certain kinds of people. 1 of whom I'm confident is Vladimir Putin, I'm for sure talking to, I really want to make it happen because I think I could do a good job of it. But the right, you know, understanding the fundamentals of translation is something I'm really interested in.
Speaker 2
16:59
So that's why I'm starting with the actual translators of like Russian literature because they understand the nuance and the beauty of the language
Speaker 3
17:07
and how
Speaker 2
17:08
it goes back and forth. But I also want to see like in speech, how can we do it in real time? So That's like a little bit of a baby project that I hope to push forward.
Speaker 2
17:19
But anyway.
Speaker 3
17:19
It's a challenging thing. So just to share, my dad actually does translation. Not professionally, he writes poetry.
Speaker 3
17:28
That was kind of always his, not a hobby, but he had a job, like a day job, but his passion was always writing poetry. And then we get to America, and he started also translating. First, he was translating English poetry to Russian. Now he also goes the other way.
Speaker 3
17:49
He kind of gained some small fame in that world anyways, because recently this poet, like Lewis Cluck, I don't know if you know of, some American poet, she was awarded the Nobel Prize for literature. And so my dad had translated 1 of her books of poetry into Russian. He was like 1 of the few, so they asked him and gave an interview to Radios Voboda, if you know what that is. And he talked about some of the intricacies of translating poetry.
Speaker 3
18:17
So that's like an extra level of difficulty, right? Because translating poetry is even more challenging than translating just, you know, interviews.
Speaker 2
18:25
Do you remember any experiences and challenges to having to do the translation that Stig got to, like something he's talked about?
Speaker 3
18:34
I mean, a lot of it, I think, is word choice, right? The way Russian is structured is first of all, quite different than the way English is structured, right? Just there is inflections in Russian and genders and they don't exist in English.
Speaker 3
18:45
That's 1 of the reasons actually why machine translation is quite difficult for English to Russian and Russian to English because they're such different languages. But then English has like a huge number of words, many more than Russian actually, I think. So it's often difficult to find the right word to convey the same emotional meaning.
Speaker 2
19:03
Yeah, Russian language, they play with words much more. So you were mentioning that Rev was kind of born out of trying to take a vertical on Upwork and then standardize it.
Speaker 3
19:17
So- We're just trying to make the freelancer marketplace idea better, right? Better for both customers and better for the freelancers themselves.
Speaker 2
19:28
Is there something else to the story of finding Rev? Like what did it take to bring it actually to life? Was there any pain points?
Speaker 3
19:38
Plenty of pain points. I mean, as often the case, it's with scaling it up, right? And in this case, you know, the scaling is kind of scaling the marketplace, so to speak, right?
Speaker 3
19:49
Rev is essentially a two-sided marketplace, right? Because there's the customers, and then there's the reverse. If there's not enough reverse, the reverse is what we call our freelancers. If there's not enough reverse, then customers have a bad experience.
Speaker 3
20:03
Takes longer to get your work done, things like that. If there's too many, then the drivers have a bad experience because they might log on to see what work is available and there's not very much work. So kind of keeping that balance is a quite challenging problem. And that's like a problem we've been working on for many years.
Speaker 3
20:23
We're still like refining our methods, right?
Speaker 1
20:25
If you can kind
Speaker 2
20:26
of talk to this gig economy idea. I did a bunch of different psychology experiments on Mechanical Turk, for example. I've asked to do different kinds of very tricky computer vision annotation on Mechanical Turk and it's connecting people in a more systematized way.
Speaker 2
20:44
I would say, you know, between task and, what would you call that, worker, is what Mechanical Turk calls it. What do you think about this world of gig economies, of there being a service that connects customers to workers in a way that's like massively distributed, like potentially scaling to, it could be scaled to like tens of thousands of people, right? Is there something interesting about that world that you can speak to?
Speaker 3
21:18
Yeah, well, we don't think of it as kind of gig economy. Like to some degree, I don't like the word gig that much, because to some degree it diminishes the work being done, right? It sounds kind of like almost amateurish.
Speaker 3
21:30
Well, maybe in like music industry, like gig is the standard term, but in work, it kind of sounds like, oh, it's, it's, it's frivolous. To us, it's, improving the nature of working from home on your own time and on your own terms, right. And kind of taking away geographical limitations and time limitations. Right.
Speaker 3
21:54
So, you know, many of our freelancers are maybe work from home moms, right. Just don't want the traditional 9 to 5 job, but they wanna make some income. And Rev kinda allows them to do that and decide exactly how much to work and when to work. Or by the same token, maybe someone wants to live the mountaintop life, right?
Speaker 3
22:19
Cabin in the woods, but they still wanna make some money. And generally that wouldn't be compatible before this new world. You kind of had to choose. But with Rev, you feel like you don't have to choose.
Speaker 2
22:31
Can you speak to like, what's the demographics, like distribution, like where do Revvers live? Is it from all over the world? Like, what is it?
Speaker 2
22:42
Do you have a sense of what's
Speaker 3
22:45
out there?
Speaker 2
22:46
It's from
Speaker 3
22:46
all over the world. Most of them are in the US, that's the majority, because most of our work is audio transcription and so you have to speak pretty good English. So the majority of them are from the US, we have people in some other of the English-speaking countries.
Speaker 3
23:03
And as far as like US, it's really all over the place. You know, for some years now, we've been doing these little meetings where the management team will go to some place and we'll try to meet Revers and, you know, pretty much wherever we go, it's pretty easy to find, you know, a large number of reverse, you know, the most recent 1 we did is in Utah. But anywhere really.
Speaker 2
23:25
Are they from all walks of life? Are these young folks, older folks?
Speaker 3
23:28
Yeah, all walks of life, really. Like I said, you know, 1 category is, you know, the work from home mom, students, you know, who want to make some extra income. There are some people who maybe have some social anxiety, so they don't want to be in the office, right?
Speaker 3
23:43
And this is 1 way for them to make a living. So it's really pretty wide variety. But like on the flip side, for example, 1 revver we were talking to was a person who had a fairly high-powered career before and was kind of like taking a break and just wanted, she was almost doing this just to explore and learn about, you know, the gig economy, quote unquote, right? So it really is a pretty wide variety of folks.
Speaker 2
24:06
Yeah, it's kind of interesting through the captioning process for me to learn about the revvers because some are clearly weirdly knowledgeable about technical concepts. Like you can tell by how good they are at like capitalizing stuff, like technical terms, like in machine learning and deep learning. Like I've used Rev to annotate, to caption the deep learning lectures or machine learning lectures I did at MIT.
Speaker 2
24:39
And it's funny, a large number of them were like, I don't know if they looked it up or were already knowledgeable, but they do a really good job at like,
Speaker 3
24:49
I don't know. They invest time into these things. They will like do research, they will Google things, you know, to kind of make sure they get it right.
Speaker 3
24:57
But to some of them, it's like, it's actually part of the enjoyment of the work. Like they'll tell us, you know, I love doing this because I get paid to watch a documentary on something, right, and I learn something while I'm transcribing, right? Pretty cool.
Speaker 2
25:10
Yeah. So what's that captioning transcription process look like for the Rever? Can you maybe speak to that to give people a sense, like how much is automated, how much is manual? What's the actual interface look like?
Speaker 2
25:24
All that kind of stuff.
Speaker 3
25:26
Yeah. So, you know, we've invested a pretty good amount of time to give like our revvers the best tools possible. You know, so typical day forever, they might log into their workspace. They'll see a list of audios that need to be transcribed.
Speaker 3
25:41
And we try to give them tools to pick specifically the ones they want to do, you know? So Maybe some people like to do longer audios or shorter audios. People have their preferences. Some people like to do audios in a particular subject or from a particular country, so we try to give people the tools to control things like that.
Speaker 3
26:01
And then when they pick what they want to do, we'll launch a specialized editor that we've built to make transcription as efficient as possible. They'll start with a speech rec draft. So, you know, we have our machine learning model for automated speech recognition. They'll start with that and then our tools are optimized to help them correct that.
Speaker 2
26:22
So it's basically a process of correction.
Speaker 3
26:25
Yeah, it depends on, you know, I would say the audio, if audio itself is pretty good, like probably like our, our podcast right now would be quite good. So the ASR would do a fairly good job. But if you imagine someone recorded a lecture, you know, in the back of a auditorium, right?
Speaker 3
26:45
Where like the speaker is really far away and there's maybe a lot of crosstalk and things like that, then maybe the ESR wouldn't do a good job. So the person might say, you know what, I'm just gonna do it from scratch.
Speaker 2
26:54
Do it from scratch, yeah. So it kind
Speaker 3
26:56
of really depends. What would you say is
Speaker 2
26:58
the speed that you can possibly get? Like what's the fastest? Can you get, is it possible to get real time or no?
Speaker 2
27:05
As you're like listening, can you write as fast as a
Speaker 3
27:09
real time would be pretty difficult. It's actually a pretty, it's not an easy job. You know, we actually encourage everyone at the company to try to be a transcriber for a day, transcriptionist for a day.
Speaker 3
27:20
And it's way harder than you might think it is, right? Because people talk fast and people have accents and all this kind of stuff. So real time is pretty difficult.
Speaker 2
27:30
Is it possible? Like there's somebody, we're probably gonna use Rev to caption this. They're listening to this right now.
Speaker 2
27:40
What do you think is the fastest you could possibly get on this right now?
Speaker 3
27:45
I think on a good audio, maybe 2 to 3XI would say. Real time.
Speaker 2
27:49
Meaning it takes 2 to 3 times longer than the actual audio of the podcast. This is so meta, I could just imagine the reverse working
Speaker 3
27:58
on this right now. Like You're way wrong.
Speaker 2
28:01
You're way wrong, this thing's way longer. But yeah, it definitely works.
Speaker 3
28:04
Or you doubted me, I could do real time.
Speaker 2
28:07
Yeah. Okay, so you mentioned ASR. Can you speak to what is ASR, automatic speech recognition? How much, like what is the gap between perfect human performance and perfect or pretty damn good ASR?
Speaker 3
28:26
Yeah, so ASR, automatic speech recognition, it's a class of machine learning problem, right? To take speech like we were talking and transform it into a sequence of words, essentially. Audio of people talking.
Speaker 3
28:38
Audio to words. And there's a variety of different approaches and techniques, which we could talk about later if you want. So we think we have pretty much the world's best ASR for this kind of speech. So there's different kinds of domains for ASR.
Speaker 3
28:56
Like 1 domain might be voice assistants, so Siri. Very different than what we're doing, right? Because Siri, there's fairly limited vocabulary. You know, you might ask Siri to play a song or, you know, word repeats or whatever.
Speaker 3
29:11
And it's very good at doing that. Very different from when we're talking in a very unstructured way. And Siri will also generally adapt to your voice and stuff like this. So for this kind of audio, we think we have the best.
Speaker 3
29:24
And our accuracy right now, I think it's maybe 14% word error rate on our test suite that we generally use to measure. So word error rate is like 1 way to measure accuracy for ASR, right? So what's 14% word error rate? So 14% means across this test suite of a variety of different audios, it would be, it would get in some way, 14% of the words wrong.
Speaker 3
29:55
14% of the words wrong. Yeah. So the way you kind of calculated this, you might add up insertions, deletions and substitutions, right? So insertions is like extra words.
Speaker 3
30:07
Deletions are words that we said, but, weren't in the transcript, right? Substitutions is you said Apple, but I said, but the ASR thought it was able, something like this. Human accuracy, most people think realistically, it's like 3%, 2% word error rate would be the max achievable. So There's still quite a gap, right?
Speaker 2
30:31
Would you say that, so YouTube, when I upload videos, often generates automatic captions. Are you, sort of from a company perspective, from a tech perspective, are you trying to beat YouTube? Google?
Speaker 2
30:45
It's a hell of a, Google, I mean, I don't know how seriously they take this task, but I imagine it's quite serious. And they, you know, Google is probably up there in terms of their teams on ASR, or just NLP, natural language processing, different technologies. So do you think you can beat Google?
Speaker 3
31:06
On this kind of stuff, yeah, we think so.
Speaker 2
31:08
Google just woke up on my phone.
Speaker 3
31:11
This is hilarious, okay. Now Google is listening, sending it back to headquarters. Who are these Rev people?
Speaker 2
31:19
But that's the goal? Yeah.
Speaker 3
31:20
I mean, we measure ourselves against like Google, Amazon, Microsoft, you know, some of the, some smaller competitors. And we use like our internal tests with it. We try to compose it of a pretty representative set of audios.
Speaker 3
31:33
Maybe it's some podcasts, some videos, some interviews, some lectures, things like that. Right. And we beat them in our own testing.
Speaker 2
31:42
And actually Rev offers automated, Like you can actually just do the automated captioning. So like, I guess it's like way cheaper, whatever it is. Whatever the rates are.
Speaker 2
31:54
Yeah, yeah. So it's a, by the way, it used to be a dollar per minute for captioning and transcription. I think it's like a dollar 15 or something like that. Dollar 25.
Speaker 2
32:03
Dollar 25. Dollar 25, no. Yeah. It's pretty cool.
Speaker 2
32:09
That was the other thing that was surprising to me. It was actually like the cheapest thing you could, I mean, I don't remember it being cheaper. You could on Upwork get cheaper, but it was clear to me that this, that's going to be really shitty. Yeah.
Speaker 2
32:24
So like, you're also competing on price. I think there were services that you can get like similar to Rev kind of feel to it, but it wasn't as automated. Like the drag and drop, the entirety of the interface, it's like the thing we're talking about. I'm such a huge fan of like frictionless, like Amazon's single buy button, whatever.
Speaker 2
32:47
Yeah, yeah, 1 click. The 1 click, That's genius right there. Like that is so important for services. Yeah.
Speaker 2
32:55
That simplicity. And I mean, Rev is almost there. I mean, there's like some, trying to think. So I think I've, I stopped using this pipeline, but Rev offers it and I like it, but it was causing me some issues on my side, which is you can connect it to like Dropbox
Speaker 3
33:20
and
Speaker 2
33:20
it generates the files in Dropbox. So like it closes the loop to where I don't have to go to Rev at all and I can download it. Sorry, I don't have to go to Rev at all and to download the files.
Speaker 2
33:34
It could just like automatically copy them.
Speaker 3
33:36
Right, you put in your Dropbox and a day later, or maybe a few hours later, depending on the giga rush, it just shows up.
Speaker 2
33:43
Yeah, I was trying to do programmatically too. Is there an API interface you can, I was trying to through like through Python to download stuff automatically, but then I realized this is the programmer in me? Like, dude, you don't need to automate everything like in life, like flawlessly, because I wasn't doing enough captions to justify to myself the time investment into automating everything perfectly.
Speaker 3
34:07
Yeah, I would say if you're doing so many interviews that your biggest roadblock is clicking on the rev download button, Now you're talking about Elon Musk levels of business. But for sure we have like a variety of ways to make it easy. You know, there's the integration.
Speaker 3
34:24
You mentioned, I think it's through a company called Zapier, which kind of can connect Dropbox to Revan, vice versa. We have an API if you want to really customize it, if you want to create the Lex Friedman CMS or whatever.
Speaker 2
34:40
But this whole thing, okay, cool. So can you speak to the ASR a little bit more? Like what does it take, like approach-wise, machine learning-wise, how hard is this problem, how do you get to the 3% error rate, like what's your vision of all of this?
Speaker 3
34:59
Yeah, well, The 3% error rate is definitely, that's the grand vision. We'll see what it takes to get there. But we believe in ASR, the biggest thing is the data.
Speaker 3
35:15
Like that's true of a lot of machine learning problems today, right? The more data you have and the higher quality of the data, the better labeled the data. That's how you get good results. And we at Rev have kind of like the best data.
Speaker 2
35:28
Like, we have... Like, you're literally... Your business model is annotating the data.
Speaker 3
35:33
Our business model is being paid to annotate the data. Being paid to annotate the data. So it's kind of like a pretty magical flywheel.
Speaker 3
35:42
And so we've kind of like ridden this flywheel to this point. And we think we're still kind of in the early stages of figuring out all the parts of the flywheel to use, because we have the final transcripts and we have the audios, and we train on that. But we in principle also have all the edits that the reverse make, right?
Speaker 2
36:07
Oh, that's interesting. How can you use that as data?
Speaker 3
36:10
We, that's something for us to figure out in the future. But, you know, we feel like we're only in the early stages, right?
Speaker 2
36:16
So the data is there, that'd be interesting, like almost like a recurrent neural net for fixing transcripts. I always remember we did segmentation annotation for driving data, so segmenting the scene, like visual data. And you can get all, so it was drawing, people were drawing polygons around different objects and so on.
Speaker 2
36:39
And it feels like, it always felt like there was a lot of information in the clicking, the sequence of clicking that people do, the fixing of the polygons that they do. Now, there's a few papers written about how to draw polygons with recurring neural nets to try to learn from the human clicking. But it was just experimental. It was 1 of those like CVPR type papers that people do like a really tiny data set.
Speaker 2
37:08
It didn't feel like people really tried to do it seriously. And I wonder if there's information in the fixing that provides deeper set of signal than just like the raw data.
Speaker 3
37:24
Of course- The intuition is for sure there must be, right? There must be. And in all kinds of signals and how long you took to make that edit and stuff like that.
Speaker 3
37:32
It's going to be up to us. That's why the next couple of years is super exciting for us.
Speaker 2
37:38
So that's what the focus is now. You mentioned Rev.ai. That's where you want to?
Speaker 3
37:43
Yeah, so Rev.ai is our way of bringing this ASR to the rest of the world. So when we started, we were human only, then we kind of created this TEMI service, I think you might've used it, which was kind of ASR for the consumer, right? So if you don't want to pay $1.25, but you want to pay, now it's 25 cents a minute, I think.
Speaker 3
38:08
And you get the transcript, the machine generated transcript, you get an editor and you can kind of fix it up yourself. Then we started using the ASR for our own human transcriptionists. And then the kind of the final step of the journey, which is, you know, we have this amazing engine. What can people build with it, right?
Speaker 3
38:28
What kind of new applications could be enabled if you have SpeedTrack that's that accurate.
Speaker 2
38:36
Do you have ideas for this or is it just providing it as a service and seeing what people come up with?
Speaker 3
38:40
It's providing it as a service and seeing what people come up with and kind of learning from what people do with it. And we have ideas of our own as well, of course, but it's a little bit like, you know, when AWS provided the building blocks, right? And they saw what people built with it and they try to make it easier to build those things, right?
Speaker 3
38:56
And we kind of hope to do the same thing.
Speaker 2
38:59
Although AWS kind of does a shitty job of like, I'm continually surprised like Mechanical Turk, for example, how shitty the interfaces. We're talking about like Rev making me feel good. Like when I first discovered Mechanical Turk, the initial idea of it was like, it made me feel like Rev does, but then the interface is like, come on.
Speaker 3
39:22
Yeah, it's horrible.
Speaker 2
39:25
Why is that so painful? Does nobody at Amazon wanna like seriously invest in it? It felt like you can make so much money if you took this effort seriously.
Speaker 2
39:37
And it feels like they have a committee of like 2 people just sitting back, like a meeting, they meet once a month, like what are we gonna do with Mechanical Turk? It's like 2 websites make me feel like this. That and craiglist.org, whatever the hell it is. Feels like it's designed in the 90s.
Speaker 3
39:55
Well, Craiglist basically hasn't been updated pretty much since the guy originally built. Do you
Speaker 2
40:00
seriously think there's a team, like how big is the team working on Mechanical Turk?
Speaker 3
40:04
I don't know, there's some team, right?
Speaker 2
40:06
I feel like there isn't, I'm skeptical.
Speaker 3
40:09
Yeah, well if nothing else they benefit from, you know, the other teams like moving things forward. Right, in a small way. But no, I know what you mean, we use mechanical Turk for a couple of things as well and it's painful.
Speaker 2
40:24
But yeah, it works.
Speaker 3
40:25
I think most people, the thing is most people don't really use the UI, right? Like so like We, for example, we use it through the API, right?
Speaker 2
40:33
But even the API documentation and so on, like it's super outdated. I don't even know what to, I mean, the same criticism, as long as we're ranting, My same criticism goes to the APIs of most of these companies, like Google, for example. The API for the different services is just, the documentation is so shitty.
Speaker 2
40:59
Like, it's not so shitty. I should actually be, I should exhibit some gratitude. Okay, let's practice some gratitude. The documentation is pretty good.
Speaker 2
41:14
Like most of the things that the API makes available is pretty good. It's just that in the sense that it's accurate, sometimes outdated, but like the degree of explanations with examples is only covering, I would say, like 50% of what's possible. And it just feels a little bit like there's a lot of natural questions that people would wanna ask that doesn't get covered. And it feels like it's almost there.
Speaker 2
41:44
Like it's such a magical thing. Like the Maps API, YouTube API, there's a
Speaker 3
41:50
bunch of stuff. I gotta imagine it's like, you know, there's probably some team at Google, right, responsible for writing this documentation. That's probably not the engineers, right?
Speaker 3
42:00
And probably this team is not, you know, where you want to be.
Speaker 2
42:04
Well, it's a weird thing. I sometimes think about this. For somebody who wants to also build the company, I think about this a lot.
Speaker 2
42:16
YouTube, the service, is 1 of the most magical, like I'm so grateful that YouTube exists. And yet, they seem to be quite clueless on so many things, like that everybody's screaming them at. Like, it feels like whatever the mechanism that you use to listen to your quote unquote customers, which is like the creators, is not very good. Like there's literally people that are like screaming, why like their new YouTube studio, for example, there's like features that were like begged for, for a really long time, like being able to upload multiple videos at the same time.
Speaker 2
43:00
That was missing for a really, really long time. Now, like there's probably things that I don't know, which is maybe for that kind of huge infrastructure, it's actually very difficult to build some of these features. But the fact that that wasn't communicated and it felt like you're not being heard. Like I remember this experience for me and it's not a pleasant experience.
Speaker 2
43:23
And it feels like the company doesn't give a damn about you. And that's something to think about. I'm not sure what that is. That might have to do with just like small groups working on these small features and these specific features.
Speaker 2
43:35
And there's no overarching like dictator type of human that says like, why the hell are we neglecting like Steve Jobs type of characters? Like there's people that we need to speak to the people that like wanna love our product and they don't. Let's fix
Speaker 3
43:52
this shit. Yeah, maybe at some point you just get so fixated on the numbers, right? And it's like, well, the numbers are pretty great, right?
Speaker 3
43:57
Like people are watching, you know? Doesn't seem to be a problem, right? Doesn't seem to be a problem. And you're not like the person that like build this thing, right, so you really care about it.
Speaker 3
44:05
You know, you're just there, you came in as a product manager, right? You got hired sometime later, your mandate is like, increase this number, like, you know, 10%, right? And you just-
Speaker 2
44:16
That's brilliantly put. Like if you, this is, okay, if there's a lesson in this, is don't reduce your company into a metric of like, how much, like you said, how much people watching the videos and so on, and like convince yourself that everything's working just because the numbers are going up. There's something, you have to have a vision.
Speaker 2
44:41
You have to want people to love your stuff because love is ultimately the beginning of a successful long-term company is that they always should love your product.
Speaker 3
44:51
You have to be like a creator and have that creator's love for your own thing, right? And you paint by these comments, right? And probably like, Apple, I think, did this generally like really well.
Speaker 3
45:03
They're well known for kind of keeping teams small even when they were big, right? And, you know, he was an engineer, like there's that book, Creative Selection, I don't know if you read it, by an Apple engineer named Ken Kosienda. It's kind of a great book actually, because unlike most of these business books where it's, you know, here's how Steve Jobs ran the company, it's more like, here's how life was like for me, you know, an engineer. Here are the projects I worked on and here what it was like to pitch Steve Jobs, you know, on like, you know, I think it was in charge of like the keyboard and the auto correction, right?
Speaker 3
45:36
And at Apple, like Steve Jobs reviewed everything. And so he was like, this is what it was like to show my demos to Steve Jobs and, you know, to change them because like Steve Jobs didn't like how, you know, the shape of the little key was off because the rounding of the corner was not quite right or something like this, but he was famously a stickler for this kind of stuff. But because the teams were small, he really owned this stuff, right? So he really cared.
Speaker 2
45:58
Yeah, Elon Musk does that similar kind of thing with Tesla, which is really interesting. There's another lesson in leadership in that is to be obsessed with the details. And like he talks to like the lowest level engineers.
Speaker 2
46:11
Okay, so we're talking about ASR. And So this is basically where I was saying, we're gonna take this like ultra seriously. And then what's the mission to try to keep pushing towards the 3%?
Speaker 3
46:26
Yeah, and kind of try to build this platform where all of your audits, all of your meetings, they're as easily accessible as your notes. So imagine all the meetings a company might have. Now that I'm no longer a programmer, and I'm a quote unquote manager, that's like my day is in meetings.
Speaker 3
46:52
And pretty often I want to see what was said, who said it, what's the context, But it's generally not really something that you can easily retrieve, right? Like imagine if all of those meetings were indexed, archived, you could go back, you could share a clip really easily.
Speaker 2
47:08
So that might change completely. Everything that's said converted to text might change completely the dynamics of what we do in this world. Especially now with remote work.
Speaker 3
47:18
Exactly, exactly.
Speaker 2
47:19
With Zoom and so on. That's fascinating to think about. I mean, for me, I care about podcasts, right?
Speaker 2
47:25
And 1 of the things that was, you know, I'm torn. I know a lot of the engineers at Spotify. So I love them very much because they dream big in terms of like, they wanna empower creators. So 1 of my hopes was with Spotify that they would use a technology like Rev or something like that to start converting everything into text and make it indexable.
Speaker 2
47:55
Like 1 of the things that sucks with podcasts is like it's hard to find stuff. Like the model is basically subscription, like you find, it's similar to what YouTube used to be like, which is you basically find a creator that you enjoy and you subscribe to them and like you just, you just kind of follow what they're doing, but the search and discovery wasn't a big part of YouTube like in the early days, but that's what currently with podcasts, is the search and discovery is non-existent. You're basically searching for the dumbest possible thing, which is like keywords in the titles of episodes.
Speaker 3
48:39
Yeah, but even aside from searching this cover, like all the time, so I listen to like a number of podcasts and there's something said and I wanna like go back to that later because I was trying to, I'm trying to remember, what do you say? Like maybe like recommend some cool product that I wanna try out, and it's basically impossible. Maybe some people have pretty good show notes, so maybe you'll get lucky and you can find it, right?
Speaker 3
48:58
But if everyone had transcripts and it was all searchable, it would be so much better.
Speaker 2
49:05
I mean, that's 1 of the things that I wanted to, I mean, 1 of the reasons we're talking today is I wanted to take this quite seriously, the rev thing. I've just been lazy. So because I'm very fortunate that a lot of people support this podcast, that there's enough money now to do a transcription and so on, it seemed clear to me, especially like CEOs and sort of like PhDs, like people write to me who are like graduate students in computer science or graduate students in whatever the heck field.
Speaker 2
49:40
It's clear that their mind, like they enjoy podcasts when they're doing laundry or whatever, but they wanna revisit the conversation in a much more rigorous way. And they really want a transcript. It's clear that they want to like analyze conversations. So many people wrote to me about a transcript for Yosha Bach conversation.
Speaker 2
50:01
I had just a bunch of conversations. And then on the Elon Musk side, like reporters want like, they wanna write a blog post about your conversation. So they wanna be able to pull stuff. And it's like, they're essentially doing on your conversation transcription privately.
Speaker 2
50:18
They're doing it for themselves and then starting to pick. But it's so much easier when you can actually do it as a reporter, just look at the transcript.
Speaker 3
50:26
Yeah, and you can like embed a little thing, you know, like into your article, right? Here's what they said. You can go listen to like this clip from the section.
Speaker 2
50:34
I'm actually trying to figure out, I'll probably on the website create like a place where the transcript goes like as a webpage so that people can reference it, like reporters can reference it and so on. I mean, most of the reporters probably have wanted to write clickbait articles that are complete falsifying, which I'm fine with. It's the way of journalism.
Speaker 2
50:56
I don't care. Like I've had this conversation with a friend of mine, a mixed martial artist, Ryan Hall. And we talked about, as I've been reading the rise and fall of the Third Reich and a bunch of books on Hitler, and we brought up Hitler and he made some kind of comment where we should be able to forgive Hitler and you know like we were talking about forgiveness and we're bringing that up as like the worst case possible thing is like even you know for people who are Holocaust survivors 1 of the ways to let go of the suffering they've been through is to forgive. And he brought up like Hitler is somebody that would potentially be the hardest thing to possibly forgive, but it might be a worthwhile pursuit psychologically, so on, blah, blah, blah.
Speaker 2
51:47
It doesn't matter. It was very eloquent, very powerful words. I think people should go back and listen
Speaker 3
51:52
to it.
Speaker 2
51:53
It's powerful. And then all these journalists, all these articles written about MMA fight, UFC fight. Right.
Speaker 3
52:00
MMA fighter loves Hitler.
Speaker 2
52:01
No, like, well, no, they were somewhat accurate. They didn't say like, loves Hitler, they said, thinks that if Hitler came back to life, we should forgive him. Like, they kind of, It's kind of accurate-ish, but the headline made it sound a lot worse than it was, but I'm fine with it.
Speaker 2
52:28
That's the way the world, I wanna almost make it easier for those journalists and make it easier for people who actually care about the conversation to go and look and see. Right,
Speaker 3
52:37
they can see it for themselves.
Speaker 2
52:38
For themselves, full context. They can go. There's something about podcasts like the audio that makes it difficult to go, to jump to a spot and to look for that, for that particular information.
Speaker 2
52:53
I think some of it, you know, I'm interested in creating like myself, experimenting with stuff. So like taking Rev and creating a transcript and then people can go to it. I do dream that like I'm not in the loop anymore. That like, you know, Spotify does it, right?
Speaker 2
53:13
Like automatically for everybody. Because ultimately that 1 click purchase needs to be there. Like, you
Speaker 3
53:21
know, I mean, like it kind of wants support from the entire ecosystem, like from the tool makers and the podcast creators, even clients, right? I mean, imagine if like, most podcast apps, you know, if it was a standard, right? Here's how you include a transcript into a podcast, right?
Speaker 3
53:38
Podcast is just an RSS feed ultimately. And actually just yesterday I saw this company called Buzzsprout, I think they're called. So they're trying to do this. They proposed a spec, an extension to their RSS format to reference transcripts in a standard way.
Speaker 3
53:57
And they're talking about like, there's 1 client dimension that will support it, But imagine like more clients support it, right? So any podcast you could go and see the transcripts, right in your like normal podcast app.
Speaker 2
54:10
Yeah, I mean, somebody, so I have somebody who works with me works with, helps with advertising, Matt, this awesome guy, He mentioned Buzzsprout to me, but he says it's really annoying because they want exclusive, they want to host the podcast.
Speaker 3
54:26
This
Speaker 2
54:26
is the problem with Spotify too. This is where I'd like to say like F Spotify. There's a magic to RSS with podcasts.
Speaker 2
54:38
It can be made available to everyone. And then there's this ecosystem of different podcast players that emerge and they compete freely. And that's a beautiful thing. That's why I go on exclusive.
Speaker 2
54:50
Like Joe Rogan went exclusive. I'm not sure if you're familiar with, he went to just Spotify. As a huge fan of Joe Rogan, I've been kind of nervous about the whole thing, but let's see. I hope that Spotify steps up.
Speaker 2
55:04
They've added video, which was very surprising that they were able to put up.
Speaker 3
55:07
So, explicit meaning you can't subscribe to the RSS feed anymore, it's only in Spotify.
Speaker 2
55:12
For now, you can until December 1st. And December 1st, everything disappears and it's Spotify only. I, you know, and Spotify gave
Speaker 1
55:24
him $100 million for that.
Speaker 2
55:25
So it's an interesting deal, but I, you know, I did some soul searching and I'm glad he's doing it, but if Spotify came to me with $100 million, I wouldn't do it. I wouldn't do, well, I have a very different relationship with money, I hate money, but I just think, I believe in the pirate radio aspect of podcasting, the freedom, and that there's something about- The open source spirit. The open source spirit, it just doesn't seem right, doesn't feel right.
Speaker 2
55:55
That said, you know, because so many people care about Joe Rogan's program, They're gonna hold Spotify's feet to the fire. Like 1 of the cool things what Joe told me is the reason he likes working with Spotify is that they They're like ride-or-die together, right? So they they want him to succeed So that's why they're not actually telling him what to do, despite what people think. They don't give them any notes on anything.
Speaker 2
56:26
They want him to succeed. And that's the cool thing about exclusivity with a platform is like, you kind of want each other to succeed. And that process can actually be very fruitful. Like YouTube, it goes back to my criticism.
Speaker 2
56:44
YouTube generally, no matter how big the creator, maybe for PewDiePie, something like that, they want you to succeed. But for the most part, from all the big creators I've spoken with, Veritasium, all those folks, you know, they get some basic assistance, but it's not like, YouTube doesn't care if you succeed or not. They have so many creators. They have like a hundred other.
Speaker 2
57:06
They don't care. So, and especially with somebody like Joe Rogan, who YouTube sees Joe Rogan not as a person who might revolutionize the nature of news and idea space and nuanced conversations. They see him as a potential person who has racist guests on, or like, you know, they see them as like a headache potentially. So, you know, a lot of people talk about this.
Speaker 2
57:37
It's a hard place to be for YouTube actually, is figuring out with the search and discovery process of how do you filter out conspiracy theories and which conspiracy theories represent dangerous untruths and which conspiracy theories are like vanilla untruths. And then even when you start having meetings and discussions about what is true or not, it starts getting weird. It starts getting weird.
Speaker 3
58:06
It's difficult these days, right? I worry more about the other side, right? Of too much, you know, too much, not censorship.
Speaker 3
58:13
Well, maybe censorship is the right word. I mean, censorship is usually government censorship, but still, yeah, putting yourself in the position of arbiter for these kinds of things, it's very difficult. And people think it's so easy, right? Like, it's like, well, you know, like no Nazis, right?
Speaker 3
58:27
What a simple principle. But you know, yes, I mean, no 1 likes Nazis. But there's like many shades of gray, like very soon after that.
Speaker 2
58:37
Yeah, and then, of course, everybody, there's some people that call our current president a Nazi, and then there's like, so you start getting, Sam Harris, I don't know if you know that is wasted, in my opinion, his conversation with Jack Dorsey. I spoke with Jack before on this podcast and we'll talk again, but Sam brought up, Sam Harris does not like Donald Trump.
Speaker 3
59:01
I do listen to his podcast. I'm familiar with his views on the matter.
Speaker 2
59:06
And he has Jack Dorsey's like, how can you not ban Donald Trump from Twitter? And so, you know, there's a set, you have that conversation. You have a conversation where some number, some significant number of people think that the current president of the United States should not be on your platform.
Speaker 2
59:24
And it's like, okay, so if that's even on the table as a conversation, then everything's on the table for conversation. And yeah, it's tough. I'm not sure where I land on it. I'm with you, I think that censorship is bad, but I also think that
Speaker 3
59:41
should- Ultimately, I just also think, you know, if you're the kind of person that's gonna be convinced, you know, by some YouTube video, you know, that, I don't know, our government's been taken over by aliens. It's unlikely that you'll be returned to sanity simply because that video is not available to you.
Omnivision Solutions Ltd
- Getting Started
- Create Transcript
- Pricing
- FAQs
- Recent Transcriptions
- Roadmap