Sam Liang of Otter.ai and Frank Ling of Groks Science
2:01AM Dec 21, 2018
Hello. Can you hear me?
Hey, can you hear me now?
Yeah, I can hear you now.
Yes. Hey, sorry about that. I have to restart such computer and stuff, but it seems to work now. Good. Good.
Hey. Nice to meet you.
Oh, I cannot hear you though.
Okay. Can you hear me now?
Okay. Sorry about that. I had to restart my computer.
The computer got stolen a couple weeks ago. He said everything.
All right. So
yeah, I was able to hear you for like 10 seconds. Then it stopped. But now it's working.
It really does. We can you can use there's a ton of phone number you could dial in.
Yeah, I just found into site.
I just saw them to Sky
you dial in through Skype.
Yeah, I'm using a phone. Oh,
Okay. So you're
Yeah, yeah, I can.
I can see you. I can hear you. There's seems to be a little bit delay.
Okay. Um, yeah, that'd be fine. I think I'm just reporting audio right now. So that that actually works out fine. And we'll have more positive between when you talk when I talk.
Yeah. Are you okay? Are you using computer audio? Or you're using a ton of phone? audios?
I'm only requiring the audio so the video is not included.
Yeah, when we talk I can we can turn off video.
Oh, did you hear me?
Okay. Great. I think it started here.
Alright, so welcome back to the program. Joining us today is our very special guest, Mr. Sam Wang will be talking about his test fiction app. honor
them was previously a lead programmer at Google under math problem. And he's been involved in various venture. Sam, thank you so much for joining us here today.
Thank you for having me here.
So I had a chance to test out otter the other day was conversations and I have in my colleague um, it's very impressive. So to start with people who have accent, sometimes it's a little harder to pick up the words, can you tell us a little bit about how the program works? And, you know, where anything the opportunities are to help has captions become easier for everyone?
Yeah, of course,
first of all, we actually don't see ourselves as a transcription company,
we see this service is much more interesting than just transcription are the transcription is a
immediate benefit people can see right away. But we see this as a collaboration tool
for teams for either
people to talk on,
we do conferences, or telephones, or when people meet in person,
because they communicated with some boys to convey ideas, to
discuss things to the debate to make a decision.
A voice is such an important way for people to communicate. But most of the information in conversation this actually last,
if you think about it, everybody actually talk and this and for several hours every day,
how much actually Can you remember?
And how much value can you get out of that three or four hours you spend in talking and listening? So
our motivation is much higher than just getting transcription prescription. It's a very first step to get the text format of the communication on top of that, there will be a lot more work to do to understand the conversation became people understand who is asking the question, who is answering the question, what are the action items they decided on? What's the summary of the
conversation. So otter right now, it's already able to do transcription really well, it handles a lot of different accents, it handles
strong background noise,
it handles reverberation, a lot of rooms, actually, there's a hole in the room.
And it handles different media channels like telephone actually, people some differently when they talk over telephone, when you use a
internet based video conferencing system, the song actually has some distortion and paired to listening to someone in person,
our AI and deep learning based system have to handle all of those,
all of those actually very easy for human ear to handle. But for computers to handle, all this variation is actually chimney challenging, we have viewed as very reliable, very accurate system to handle this task. So using order, you can get the meeting notes automatically, you can actually share the meeting notes with each other,
I'm actually doing it right now using order on my computer, I can share the real time transcription with you actually
paste this URL into zoom, if you click on that, and they will open a web page that will show the transcript. So this help you actually communicate in real time because
sometimes people can distracted the didn't hear a number very clearly, instead of asking the question, sorry, can you repeat that you can also just look at the transcript and quickly get the idea
the sack or even more interesting one, you have a team you know, in enterprises, you can easily have 10 or 20 people attending a meeting, people are multitasking. And not everybody is you know, paying attention all the time. But
you can always look at the transcript and quickly get what people just discussed. And then you can jump in and right and pretend you you did pay attention but you actually get access to all the information it's almost like using your you know, this is where DVR digital video recorder is so powerful. You can rewind and you can quickly replay the and artwork can serve that purpose as well. Because you can
go back to five minutes ago and reveal what did people say.
So it has real time value is also has value after the meeting is finished. Again, as I mentioned, the number one benefit people see immediately is know taking it helps the member things helps search
for a busy professional, they have a lot of meetings every day, they talked to the product managers to talk to the marketing team. They talk to the engineers,
you can easily spend several hours in all kinds of meetings, especially for higher level managers, directors will be peace,
there's actually a really good productivity system to help them.
So definitely, we see this much more powerful than the traditional transcription
and the user base is much broader as well. Traditionally, only a small
niche of people actually need to get transcription reporters or core reporters, majority of the people actually don't even know they need transcription.
But honor the way we design it. It's very easy to use, you can sign up and started using it within 10 seconds. works on iPhone works on Android works inside the browser actually
in on your computer. If you just go to otter.ai.
You don't need to download anything the same way you use Gmail inside a browser. UK, use otter inside the browser.
You can handle all the video conferences with otter on your computer. Whether you use Skype whether you use WebEx or zoom or Google Hangouts actually helps you get that transcript.
Sorry, I cannot hear you now.
So I really loved her icon. How did you get the idea you can order or your program?
because we think this is
can be used by a very broad
We want to have a icon or logo that's very friendly,
honor is a very adorable animal,
you probably can see a lot of the videos in YouTube and on the internet that shows otters actually hold hands in the water and the mother other kid carry their baby on their tummy and the floating on the water.
They're very friendly.
They are a lot of people actually don't know otters are one of the smartest animals in the world. They have very high IQ,
they can learn a lot of complicated skills.
I've seen orders people train harder to play basketball, you water,
they can use crews, others, I also have very good memory, they remember things really well. So so we picked otter as the name of the product.
And people really like it.
So now this is otter with artificial intelligence. So you can go to otter.ai your browser to use it or download the app on your iPhone or Android.
So you weren't previously an engineer at Google. Um, what lessons can did you learn from the company that now applies to things startup are exploring new innovation?
Yeah, that'd be a lot of lessons on you know, I did my PhD in Stanford in computer distributed systems.
My future advisor, David Sheraton. He is a very
well known professor in the meantime, he's also an entrepreneur, he builds a lot of successful startups,
he actually wrote the very first check to Larry Page and Sergey to start a Google.
So at Google, I was a lead on Google Map location service for four years, we viewed a service that
useful for, you know, hundreds of millions of people.
So I to Google and learn to understand the scale of data scale,
complicated systems, and how the how, you know, provide service for hundreds of millions of people.
And whatever you do, you are impacting so many people.
So you get to learn to think big.
And now also, in terms of innovation, you're not thinking about making tiny improvement, you thinking about making two x five x 10 x equal.
after four years at Google, I decided to quit. And the work on my own startup,
it wasn't easy. The first year I had zero dollar salary at Euro dollar funding. So I had to just work on,
building a prototype in my backyard
until later, you know, I got an offer to move into an office on
sun Hill Road in the VCs office. But
definitely, and, you know, made a lot of good friends at Google, you know, they're super smart people.
now, our own new team actually have a number of people from Google originally. And,
you know, this is are not the best engineers, the best scientists in the world, you know, they build the speech recognition system that's so accurate, so fast,
really proud of our team.
Great. So for people who are very
interested in trying your app,
where should they go, and what is the business model.
So I mentioned the auto designed to be very friendly, very easy to use. So people can just go to
App Store and download the app. otter Oracle to use Android, you can go to google play store to install the app,
because sign up very easily. You can also use it in your web browser on your laptop, without downloading anything you can use other as well.
So business model wise,
we have a freemium model, meaning that every user we give them 10 hours usage for free every month, you can already using it without paying any money,
you can use it for your meetings, for your phone calls for your video conferences.
Students can use it for lectures,
we actually have a lot of users from universities to use order to take lecture notes.
And for people who want to use some events that features
and also needs to use more than 10 hours, there is a $9 99 cents subscription fee,
or the user can pay $80 for annual subscription for education. University students, we keep them 50% off
enterprise version on top of the existing otter
enterprises and could be a small team could be 10 people start up,
they can easily adopt otter for their own meetings. So our models will be a lot like Dropbox and Slack,
a lot of small teams can adopt it quickly. It could be self service, they can, you know, sign up and
just invite 10 people in your team into the office system and they can start using right away
So you know, obviously you're the first targets of the English language but most of the roles speak other languages are there plans to take water to for example Chinese or French or Russian
yeah longer term the Stephanie the plan
there's so many different languages to work with. And
and also automatic translation could be provided between different languages.
Short term we like to focus on English market first to better understand the use cases provide no more interesting features to help people collaborate better save time and also provide more analytics NLP understanding of the conversation. Again, as I mentioned earlier, this is not just a transcription although transcriptions one part of the service. But
top of that there will be much more interesting services to help people
communicate better, collaborate better
extract insights from the conversations that we have large companies active who want to use order to analyze their
meeting recordings in the last 20 years to understand how they spend their time in meetings, how they reach certain decisions. So,
you know, again, much more interesting than just
finally, I guess we're running a little of time, I just had a technical question. So what's nailing the innovation and then mostly the hardware side or the software or new algorithms
for us, we're focused on software and you ever them, this is a deep learning based AI system.
So we have lots of the deep learning models, a lot of training face on a huge quantity of data,
the more people use it, the AI can learn more, we also allow users to correct some errors we made because this is definitely not 100% perfect.
When we made mistake, if the user corrected our system will learn from it. So the more people use it, the more people correct errors
system can improve itself, this is going to be a virtuous cycle.
This is you know, again, the power of AI, the more you use it the
better you can learn from the feedback
and it will keep improving itself it's like you probably heard about alpha go build by Google they feel they beat the past ago player in the world
and then you know, the different versions of the goal, Aqua some trust to play each other, and they get a better and better
now, they, you know, even stronger than the one you know, that beat the human champion
the same here in for our anchors, and the more you use it, the better it can enhance. Yeah, it's, it's very exciting to
great Wait, thank you so much for your inspiring ideas on your app, or any last words you like that about yourself or other
we definitely would love to hear from all users love to hear their feedback and understand what users cases they use it in, in meetings, phone calls, with the conferences, or some people use it for their personal life. Like when the visit doctor's office,
you know, high school students, we see them use it as well, you know, parents, they use it in their parent teacher meetings. So wherever voice happens, honored, can be used. But we love to hear user stories and, you know, love to get some help from our users and help us evangelize, honor in get, you know, get more people to benefit from the new innovation. And this is in some sense, we really think it is going to change the world change the way we communicate, change the way we
It will also you know, is extension of human brain.
Sam, thank you so much for taking time to talk with me today.
Great. Thank you, friend.
So I stopped recording right now I'll do some editing. And then I'll aim to have this up calling next Wednesday
May or the one after that will be a little bit about more about the graph science and who your audience is
yeah, but they'll have,
right. So I'm the trigger for and we started at UC Berkeley when Charles My partner and I were in grad school. And after we graduated, we still continue. So we've been doing about one episode
every week. And so right now, we're syndicated about 20 stations around the world, the main ones University of Chicago and University of Louisiana to lane.
And then we're also on the NSF network PRF which is a distribution system. So unfortunate. I don't know how many people actually listen. But the
audience the University Chicago at least around 20,000 at any given moment,
you know, in total of 20 stations that syndicate so some of them are in Australia, some of them are
So it depends on what the station one,
you know, science programs tend not to have high priority. So unless you're around an area where the listeners are more educated, it's hard again, audience that's why we tend to approach universities or places where when people have higher international activities,
a lot of university students
army, a lot of university students,
students are probably main audience.
So I don't think we have to say the business community. It's a little bit more on the technical, um, you know, science to the science people who are a love for science and want to know more about technology
rather than the people
who are trying to implement or commercializing
Yeah, we are we
Yeah, I studied computer science myself, so I can
definitely very interesting or the scientific and algorithmic problems and engineering problems.
Right. Right. So,
so you graduated from how, why you you live in Japan now? And what other business do you this your full time job to
radio is actually more of a hobby I I focus mostly on energy and climate Ching. So I'm involved with trying to raise attention and funding for marriage energy technology.
One of the companies is called a parent's, which is now run by toward sales. He was the former building manager at Google.
That's one example. The other is um, we work with terrestrial energy. It's a Canadian nuclear company. So they have a new technology that uses moment Morton Salt to deliver the heat from the nuclear plant to the turbine.
So we're looking for next generation nuclear
technologies that don't image here to mention, but also don't have the risk of a meltdown like Fukushima
I see. Um, yeah, so if you have an interest to explore Japan, I do know, a couple of people.
One of my co authors is Norio Murakami to us the former chair of Google Japan.
Right now he's retired he he does some advice
advising for SoftBank.
Yeah we a lot of Japanese companies are interested in our system although we don't support Japanese right now.
So definitely we want to focus on English speaking first any companies you work with if they're the main language is English you know you can recommend the otter for them because we see this is actually a great collaboration and great productivity system for you know
to improve their productivity
and you can actually even publish your podcast with offer as well because
when you publish podcast traditionally only have audio but now we actually have both audio and transferred benefit of a transcript is actually can do SEO on it because the Google can index your transcript
and this is actually
one example actually I I had this podcast
recording with a podcaster in Australia and he acted pop publish it using otter look at that send you the link you scroll down near the bottom there's one iframe there that just shows
the otter transcript and audio you can play it inside the frame and it's actually
no synchronize the audio and the
the transfer word by word
of course it for the pain the probably want to do a little bit correction because the transfer is definitely not 100% perfect. You can you can correct the wrong words before you publish publisher.
Okay, great. I asked. You have to run to another call right now. But I just want to say thank you so much for taking the time
with them about your project and also let you know and this is up and just any opportunity to explore business development in Japan. We can talk about that later
now. We live in the South Bay Los Altos. That's our office.
Okay, but I know my wife, my wife got her master's degree from Cal in biostatistics.
So we love kill.
All right. Hey, thank you so much. Thank you. Take care. Yeah. All right.