News from Industry

Our AI in RTC report is just about ready. Here are all of its price points.

If you aren’t interested in AI and RTC, then move on  – this one isn’t for you.

Still here?

Good.

In the past several months I’ve been adding into my daily activities the creation of a new report – one about AI in RTC.

It has taken its toll – I’ve slept a bit less. Read a bit less. Turned down and postponed a few clients. All in order to get this project going. I’ve partnered with Chad Hart on it, one of my partners in crime at Kranky Geek and a fellow consultant.

We wanted to work on something new and interesting and this seemed to be the right thing to do.

After countless hours in interviews with vendors and suppliers in this space, discussions we had with one another and time spent just looking at the ceiling of my office and thinking, I can say that we’re almost ready with the report. Most of it is already written, and what is left will be completed really soon.

What will you find in this report?

• An introduction to machine learning and artificial intelligence. A high level one, which should be suitable for people who are less conversant in it
• Speech Analytics. A thorough chapter looking at how speech analytics is used in real time communications, including use cases, vendors and a lot more. I’d say the majority of the writing is here, as most of the focus of our industry is here
• Voice Bots. While a lot is said about chatbots, we decided to skip them (it would have de-focused us) and instead look at the domain of voice bots. Think Google Duplex, but for the enterprise
• Computer Vision. You probably saw just like me how autonomous driving is taking out the life out of computer vision elsewhere. That said, there are still vendors and places in RTC where you can find computer vision, which is what’s in this chapter of our report
• Cost and Quality Optimization. That’s the silent participant in every VoIP session you have. And it is slowly moving towards AI as well. We’ve found those who use it today and talked to those who don’t, trying to figure out both sides of the equation
• Survey summary. Remember that online survey? We’re still collecting the final responses, so be sure to fill it out if you haven’t. That’s where we will be writing our analysis if the responses we’ve received
• Other things?
  • The introductory ebook on AI in RTC (still not written), that is also given for free to ALL those filling the online survey
  • Glossary of terms related to RTC
  • A powerpoint deck of all the illustrations from the report

Where can you learn more about the report?

Three places:

• Kranky Geek website, under research
• The report brochure (PDF)
• By emailing us

How much does it cost?

Publication date is scheduled to end of July. We might miss it by a few days due to editing and some last minute changes.

• Prepublication price: $1,170 (available until publication)
• Launch discount: $1,950 (available until September 7)
• Official price: $2,950

We’re allowing payment via PayPal and wire transfer inside the US. We don’t have any digital shopping cart, as this is a first for us through Kranky Geek Research. It also means we’re treating each and every purchaser as royalty

Why wait for the price to raise? Join those who’ve already purchased at our discounted prepublication price. Interested? Just email us.

 

The post AI in RTC: Final Price Points and End of Prepublication Discount appeared first on BlogGeek.me.

Just a short note to point that a minimal proxy configuration file has been recently pushed in misc/examples/mixed, respectively:

•  https://github.com/kamailio/kamailio/blob/master/misc/examples/mixed/kamailio-minimal-proxy.cfg

It could be a good starting point, especially for testing new modules that are either developed or when starting to play with a new module, because it provides a simple SIP proxy that takes care of routing requests within dialog, but leaves out components usually enabled in production, such as authentication, nat traversal, presence, … in this way making testing on local system/network easier.Enjoy!Thanks for flying Kamailio!

Autonomous cars are sucking all the oxygen out of video AI in real time comms. Talent is focusing elsewhere

I went to the data science summit in Israel a month or so back. It was an interesting day. But somehow, I had to make sure to dodge all the boring autonomous cars sessions .they just weren’t meant for me, as I was wondering around, trying to figure out where machine learning and AI fit in RTC (you do remember I am working on a report on this – right?).

After countless of interviews done this past month, along with my partner in crime here, Chad Hart, I can say that I now know a lot more about this topic. We’ve mapped the industry in and out. Talking to technology vendors, open source projects, suppliers, consumers, you name it.

There were two interesting themes that relate to the use of AI in video – again – focus is on real time communications:

1. There’s a lot less expertise to go around in the industry, where the industry is real time comms and not machine learning or computer vision in general
2. The industry’s standards and capabilities seem higher and better than what we see in RTC today

Guess what – we’re about to incorporate the responses we got on our web survey on AI in RTC into the report. If you fill it, you’ll get our upcoming “Introduction to AI in RTC ebook” and a chance to win on of 5 $100 Amazon gift cards – along with our appreciation of helping us out. Why wait?

Fill out the web survey

Knowledge in AI is lacking

In broad strokes, when you want to do something with AI, you’ll need to either source it from other vendors or build it on your own.

As an example, you can just use Amazon Rekognition to handle object classification, and then you don’t need a lot of in-house expertise.

The savvy vendors will have people handling machine learning and AI internally as well. Being in the build category, means you need 3 types of skills:

1. Data scientists – people who can look at hoards of data, check out different algorithms and decide on what works best – what pieces of data to look at and what model to build
2. Data engineers – these are the devops of this field. They are there to connect the dots of the different elements in the system and build a kind of a pipeline where data gets processed and handled. They don’t need to know the details of algorithms, but they do need to know the jargon and concepts
3. Product managers – these are the guys who need to decide what to do. Without them, engineers will play without any focus or oversight, wasting time and resources instead of working towards value creation. These product managers need to know a thing or two about data science, machine learning and how it works

Data scientists are the hardest to find and retain. In one of our interviews, we were told that the company in question had to train their internal workforce for machine learning because it was impossible to hire experience in the valley – Google, Apple, Facebook and Amazon are the main recruiters for that position and they are too competitive in what they offer employees.

Data engineers are probably easier to find and train, but what is it you need them to do exactly?

And then there’s product managers. I am not even sure there’s any training program specifically for product managers who need to work in this space. I know I am still learning what that means exactly. Part of it by asking through our current research how do vendors end up adding AI into their products. The answers vary and are quite interesting.

Anyways – lots of hype. Less in the way of real skills out there you can hire for the job.

Autonomous driving is where computer vision is today

If you follow the general technology media out there, then there are 3 things that bubble up to the surface these days when it comes to AI:

1. AI and job displacement
2. The end of privacy (coupled with fake news in some ways)
3. Autonomous cars

The third one is a very distinct use case. And it is the one that is probably eating away a lot of the talent when it comes to computer vision. The industry as a whole is interested for some reasons to take a stab at making cars drive on their own. This is quite a challenge, and it is probably why so many researchers are flocking towards it. A lot of the data being processed in order to get us there is visual data.

Vision in autonomous cars cannot be understated. This ABC News clip of the recent Uber accident drives that point home. Look at these few seconds explaining things:

“These vehicles are trained to see pedestrians, to see cyclists, to see redlights. So it’s really unclear what went wrong here”

And then you ask a data scientist to deal withboring video meeting recordings to do whatever it is we need to do in real time communications with AI. Not enough fame in it as opposed to self driving cars. Not enough of a good story to tell your friends when you meet them after work.

Computer vision in video meetings is nascent

Then there’s the actual tidbit of what we do with AI in computer vision versus what we do with AI in video meetings.

I’d like to break this down into a table:

Computer vision Video meeting AI

• Count faces/people
• Speaker identification
• Facial recognition
• Gesture control
• Emotion detection

• Auto-frame participants

Why is this difference? Two main reasons:

1. Video meetings are real time in nature and limited in the available compute power. There’s more on that in our upcoming report. But the end result is that adopting the latest and greatest that computer vision has to offer isn’t trivial
2. We haven’t figured out as an industry where’s the ROI in most of the computer vision capabilities when it comes to video meetings – there are lower hanging fruit these days in the form of transcription, translation and what you can do with speech

As we move forward, companies will start figuring this one out – deciding how data pipeline for computer vision need to look like in video meetings AND decide what use cases are best addressed with computer vision.

Where are we headed?

The communication market is changing. We are seeing tremendous shifts in our market – cloud and APIs are major contributors to this. Adding AI into the mix means change is ahead of us for years to come.

On my end, I am adding ML/AI expertise to the things I consult about, with the usual focus of communications in mind. If you want to take the first step into understanding where AI in RTC is headed, check out our upcoming report – there’s a discount associated with purchasing it before it gets published:

AI in RTC

You can download our report prospectus here.

The post Autonomous Cars Are Killing Video AI in RTC appeared first on BlogGeek.me.

Not long time left until ClueCon 2018 starts in Chicago (July 23-26), the annual conference for RTC developers organised by the FreeSwtich team!Kamailio project is looking forward to meeting again the VoIP community and well known RTC developers around the world. Two members of Kamailio Admin Team will give presentations during the event:

• Daniel-Constantin Mierla, on Tuesday, July 24, at 12:00pm – Elasticity of VoIP Platforms with Kamailio
• Fred Posner, on Thursday, July 26, at 01:30pm – Kamailio as a SIP Edge Proxy

See the full agenda at:

• https://www.cluecon.com/schedule.html

While you are sure you will get the latest updates about FreeSwitch and SignalWire from Anthony Mineseale II, Mike Jerris, Brian West and the rest of the team, and besides Kamailio, the event covers many other open source projects, such as Asterisk, Kazoo PBX, Fusion PBX, as well as it includes talks from renowned people in the RTC space, among them: prof. Carol Davids from IIT Chicago, James Tagg, James Body, Dan Jenkins, Simon Woodhead and Chad Hart.It is one of those events during the year than one should not miss, try to hurry in order get your ticket if you haven’t done it yet and plan to attend!Looking forward to the event!Thanks for flying Kamailio!

WebRTC H.264 hardware acceleration is no guarantee for anything. Not even for hardware acceleration.

There was a big war going on when it came to the video codec in WebRTC. Should we all be using VP8 or should we be using H.264? A lot of digital ink was spilled on this topic (here as well as in other places). The final decision that was made?

Both VP8 and H.264 became mandatory to implement by browsers.

So… which of these video codecs should you use in your application? Here’s a free mini video course to help you decide.

Enroll to free course

Fast forward to today, and you have this interesting conundrum:

• Chrome, Firefox and Edge implement VP8 and H.264
• Safari implements H.264. No VP8

Leaving aside the question of what mandatory really means in English (leaving it here for the good people at Apple to review), that makes only a fraction of the whole story.

There are reasons why one would like to use VP8:

1. It has been there from the start, so its implementation is highly optimized already
2. Royalty free, so no need to deal with patents and payments and whatnot. I know there’s FUD around patents in VP8, but for the most part, 100% of the industry is treating it as free
3. It nicely supports simulcast, so quite friendly to video group calling scenarios

There are reasons why one would like to use H.264:

1. You already have H.264 equipment, so don’t want to transcode – be it cameras, video conferencing gear or the need to broadcast via HLS or RTMP
2. You want to support Safari
3. You want to leverage hardware based encoding and decoding to increase battery life on your mobile devices

I want to open up the challenges here. Especially in leveraging hardware based encoding in WebRTC H.264 implementations. Before we dive into them though, there’s one more thing I want to make clear:

You can use a mobile app with VP8 (or H.264) on iOS devices.

The fact that Apple decided NOT to implement VP8, doesn’t bar your own mobile app from supporting it.

WebRTC H.264 Challenges

Before you decide going for a WebRTC H.264 implementation, you should need to take into consideration a few of the challenges associated with it.

I want to start by explaining one thing about video codecs – they come with multiple features, knobs, capabilities, configurations and profiles. These additional doozies are there to improve the final quality of the video, but they aren’t always there. To use them, BOTH the encoder and the decode need to support them, which where a lot of the problems you’ll be facing stem from.

#1 – You might not have access to a hardware implementation of H.264

In the past, developers had no access to the H.264 codec on iOS. You could only get it to record a file or playback one. Not use it to stream media in real time. This has changed and now that’s possible.

But there’s also Android to contend with. And in Android, you’re living in the wild wild west and not the world wide web.

It would be safe to say that all modern Android devices today have H.264 encoder and decoder available in hardware acceleration, which is great. But do you have access to it?

The illustration above shows the value chain of the hardware acceleration. Who’s in charge of exposing that API to you as a developer?

The silicon designer? The silicon manufacturer? The one who built the hardware acceleration component and licensed it to the chipset vendor? Maybe the handset manufacturer? Or is it Google?

The answer is all of them and none of them.

WebRTC is a corner case of a niche of a capability inside the device. No one cares about it enough to make sure it works out of the factory gate. Which is why in some of the devices, you won’t have access to the hardware acceleration for H.264 and will be left to deal with a software implementation.

Which brings us to the next challenge:

#2 – Software implementations of H.264 encoders might require royalty payments

Since you will be needing a software implementation of H.264, you might end up needing to pay royalties for using this codec.

I know there’s this thing called OpenH264. I am not a lawyer, though my understanding is that you can’t really compile it on your own if you want to keep it “open” in the sense of no royalty payments. And you’ll probably need to compile it or link it with your code statically to work.

This being the case, tread carefully here.

Oh, and if you’re using a 3rd party CPaaS, you might want to ask that vendor if he is taking care of that royalty payment for you – my guess is that he isn’t.

#3 – Simulcast isn’t really supported. At least not everywhere

Simulcast is how most of us do group video calls these days. At least until SVC becomes more widely available.

What simulcast does is allows devices to send multiple resolutions/bitrates of the same video towards the server. This removes the need of an SFU to transcode media and at the same time, let the SFU offer the most suitable experience for each participant without resorting to lowest common denominator type of strategies.

The problem is that simulcast in H.264 isn’t available yet in any of the web browsers. It is coming to Chrome, but that’s about it for now. And even when it will be, there’s no guarantee that Apple will be so kind as to add it to Safari.

It is better than nothing, though not as good as VP8 simulcast support today.

#4 – H.264 hardware implementations aren’t always compatible with WebRTC

Here’s the kicker – I learned this one last month, from a thread in discuss-webrtc – the implementation requirements of H.264 in WebRTC are such that it isn’t always easy to use hardware acceleration even if and when it is available.

Read this from that thread:

Remember to differentiate between the encoder and the decoder.

The Chrome software encoder is OpenH264 – https://github.com/cisco/openh264

Contributions are welcome, but the encoder currently doesn’t support either High or Main (or even full Baseline), according to the README file.

Hardware encoders vary greatly in their capabilities.

Harald Alvestrand from Google offers here a few interesting statements. Let me translate them for you:

• H.264 encoders and decoders are different kinds of pain. You need to solve the problem of each of these separately (more about that later)
• Chrome’s encoder is based on Cisco’s OpenH264 project, which means this is what Google spend the most time testing against when it looks at WebRTC H.264 implementations. Here’s an illustration of what that means:
• The econder’s implementation of OpenH264 isn’t really High profile or Main profile or even Baseline profile. It just implements something in-between that fits well into real time communications
• And if you decide not to use it and use a hardware encoder, then be sure to check what that encoder is capable of, as this is the wild wild west as we said, so even if the encoder is accessible, it is going to be like a box of chocolate – you never know what they’re going to support

And then comes this nice reply from the good guys at Fuze:

@Harald: we’ve actually been facing issues related to the different profiles support with OpenH264 and the hardware encoders. Wouldn’t it make more sense for Chrome to only offer profiles supported by both? Here’s the bad corner case we hit: we were accidentally picking a profile only supported by the hardware encoder on Mac. As a result, when Chrome detected CPU issues for instance, it would try to reduce quality to a level not supported by the hardware encoder which actually led to a fallback to the software encoder… which didn’t support the profile. There didn’t seem to be a good way to handle this scenario as the other side would just stop receiving anything.

If I may translate this one as well for your entertainment:

• You pick a profile for the encoder which might not be available in the decoder. And Chrome doesn’t seem to be doing the matchmaking here (not sure if that true and if Chrome can even do that if it really wanted to)
• Mac’s hardware acceleration for the encoder of H.264, as any other Apple product, has its very own configuration to it, which is supported only by it. But somehow, it doesn’t at some point which kills off the ability to even use that configuration when you try to fallback to software
• This is one edge case, but there are probably more like it lurking around

So. Got hardware encoder and/or decoder. Might not be able to use it.

#5 – For now, H.264 video quality is… lower than VP8

That implementation of H.264 in WebRTC? It isn’t as good as the VP8 one. At least not in Chrome.

I’ve taken testRTC for a spin on this one, running AppRTC with it. Once with VP8 and another time with H.264. Here’s what I got:

VP8

Bitrate:

Framerate:

H.264

Bitrate:

Framerate:

This is for the same scenario running on the same machines encoding the same raw video. The outgoing bitrate variance for VP8 is 0.115 while it is 0.157 for H.264 (the lower the better). Not such a big difference. The framerate of H.264 seems to be somewhat lower at times.

I tried out our new scoring system in testRTC that is available in beta on both these test runs, and got these numbers:

The 9.0 score was given to the VP8 test run while H.264 got an 8.8 score.

There’s a bit of a difference with how stable VP8’s implementation is versus the H.264 one. It isn’t that Cisco’s H.264 code is bad. It might just be that the way it got integrated into WebRTC isn’t as optimized as the VP8’s integration.

Then there’s this from the same discuss-webrtc thread:

We tried h264 baseline at 6mbps. The problem we ran into is the bitrate drastically jumped all over the place.

I am not sure if this relates to the fact that it is H.264 or just to trying to use WebRTC at such high bitrates, or the machine or something else entirely. But the encoder here is suspect as well.

I also have a feeling that Google’s own telemetry and stats about the video codecs being used will point to VP8 having a larger portion of ongoing WebRTC sessions.

#6 – The future lies in AV1

After VP8 and H.264 there’s VP9 and H.265 respectively.

H.265 is nowhere to be found in WebRTC, and I can’t see it getting there.

And then there’s AV1, which includes as its founding members Apple, Google, Microsoft and Mozilla (who all happen to be the companies behind the major web browsers).

The best trajectory to video codecs in WebRTC will look something like this:

Why doesn’t this happen in VP8?

It does. To some extent. But a lot less.

The challenges in VP8 are limited as it is mostly software based, with a single main implementation to baseline against – the one coming from Google directly. Which happens to be the one used by Chrome’s WebRTC as well.

Since everyone work against the same codebase, using the same bitstreams and software to test against, you don’t see the same set of headaches.

There’s also the limitation of available hardware acceleration for VP8, which ends up being an advantage here – hardware acceleration is hard to upgrade. Software is easy. Especially if it gets automatically upgraded every 6-8 weeks like Chrome does.

Hardware beats software at speed and performance. But software beats hardware on flexibility and agility. Every. Day. of. The. Week.

What’s Next?

The current situation isn’t a healthy one, but it is all we’ve got to work with.

I am not advocating against H.264, just against using it blindingly.

How the future will unfold depends greatly on the progress made in AV1 as well as the steps Apple will be taking with WebRTC and their decisions of the video codecs to incorporate into Webkit, Safari and the iOS ecosystem.

Whatever you end up deciding to go with, make sure you do it with your eyes wide open.

So… which of these video codecs should you use in your application? Here’s a free mini video course to help you decide.

Enroll to free course

 

The post The Challenging Path to WebRTC H.264 Video Codec Hardware Support appeared first on BlogGeek.me.

Parallax, or eye contact in video conferencing is a problem that should be solved, and AI is probably how we end up solving it.

I’ve been working at a video conferencing company about 20 years ago. Since then a lot have changed:

• Resolutions and image quality have increased dramatically
• Systems migrated from on prem to the cloud
• Our focus changed from large room systems, to mobile, to desktop and now to huddle rooms
• We went from designed hardware to running it all on commodity hardware
• And now we’re going after commodity software with the help of WebRTC

One thing hasn’t really changed in all that time.

I still see straight into your nose or straight at your forehead. I can never seem to be able to look you in the eye. When I do, it ends up being me gazing straight at my camera, which is unnatural for me either.

The reason for this is known as the parallax problem in video conferencing. Parallax. What a great word.

If you believe Wikipedia, then “Parallax is a displacement or difference in the apparent position of an object viewed along two different lines of sight, and is measured by the angle or semi-angle of inclination between those two lines.”

A mouthful. Let me illustrate the problem:

What happens here is that as I watch the eyes of the person on the screen, my camera is capturing me. But I am not looking at my camera. I am looking at an angle above or beyond it. And with a group call with a couple of people in it in Hollywood squares, who should I be looking at anyway?

So you end up with either my nose.

Or my forehead.

What we really want/need is to have that camera right behind the eyes of the person we’re looking at on our display – be it a smartphone, laptop, desktop or room system.

Over the years, the notion was to “ignore” this problem as it is too hard to solve. The solution to it usually required the use of mirrors and an increase in the space the display needed.

Here’s an example from a failed kickstarter project that wanted to solve this for tablets – the eTeleporter:

The result is usually cumbersome and expensive. Which is why it never caught on.

There are those who suggest tilting the monitor. This may work well for static devices in meeting rooms, but then again, who would do the work needed, and would the same angle work on every room size and setup?

When I worked years ago at a video conferencing company, we had a European research project we participated in that included 3D imaging, 3D displays, telepresence and a few high end cameras. The idea was to create a better telepresence experience that got eye contact properly as well. It never saw the light of day.

Today, multiple cameras and depth sensors just might work.

Let’s first take it to the extreme. Think of Intel True View. Pepper a stadium with enough cameras, and you can decide to synthetically re-create any scene from that football game.

Since we’re not going to have 20+ 5K cameras in our meeting rooms, we will need to make do with one. Or two. And some depth information. Gleaned via a sensor, dual camera contraption or just by using machine learning.

Which is where two recent advancements give a clue to where we’re headed:

1. Apple Memoji (and earlier Bitmoji). iPhone X can 3D scan your face and recognize facial movements and expressions
2. Facebook can now open eyes in selfie images with the help of AI

The idea? Analyze and “map” what the camera sees, and then tweak it a bit to fit the need. You won’t be getting the real, raw image, but what you’ll get will be eye contact.

Back to AI in RTC

In our interviews this past month we’ve been talking to many vendors who make use of machine learning and AI in their real time communication products. We’ve doubled down on computer vision in the last week or two, trying to understand where is the technology today – what’s in production and what’s coming in the next release or two.

Nothing I’ve seen was about eye contact, and computer vision in real time communication is still quite nascent, solving simpler problems. But you do see the steps taken towards that end game, just not from the video communication players yet.

Interested in AI and RTC? Check out our upcoming report and be sure to assist us with our web survey (there’s an ebook you’ll receive and 5 $100 Amazon Gift cards we will raffle).

The post Can AI and Computer Vision solve the video conferencing eye contact problem? appeared first on BlogGeek.me.

Is it machine learning or artificial intelligence? It ends up depending who you ask and what is it you care about.

There are multiple ways to think and look at machine learning and artificial intelligence. And just like any other hyped technologies, people seem to mix the two and use them interchangeably.

I’ll let you in on a little secret: we’re doing the same with our upcoming AI in RTC report.

Want to help us with our research AND get a free ebook AND have a chance to win one of five $100 Amazon gift cards?

Fill out our AI in RTC survey

We could have just as easily use the title “ML in RTC” instead of “AI in RTC”. The way we’d approach and cover the space and end up writing this market research would be… the same – in both cases.

Why?

1. I’ve never been a stickler to such details, especially when so many are mixing them up
   1. This is the same as having VoIP, Convergence, UC and now Teams mean the exact same things – just slightly differently
   2. Or why WebRTC is both a standard specification (almost at least) and an open source project implementing an approximation of that standard specification
   3. And it is why people mix between ML and AI. The distinctions aren’t big enough for most of the population to care – or understand
2. Marketing
   1. Whenever a new technology or term becomes interesting and gets hyped, overzealous marketing and sales people would start using it and abusing it
   2. Which is what we see with this whole AI thing that is just everywhere now
   3. So why not us with our new report about AI in RTC?

Which brings me to this article.

Machine Learning and Artificial Intelligence are somewhat different from one another. The problem is to decide what that difference is.

Here are 4 ways to think about ML and AI:

#1 – ML = AI

Let’s start with the easiest one: ML is AI. There’s no difference between the two and they can be used interchangeably.

This is the viewpoint of the marketer, and today, of the market itself.

When everyone talks about AI, you can’t not talk about AI. Even if what you do is just ML. Or BigData. Or analytics. Or… whatever. Just say you’re doing AI. It is good for the health of your stock price.

While at it, make sure to say you’re doing AI in an ICO cryptocurrency fashion. What can go wrong?

Someone tells you he is doing AI? Assume ML, and ask for more information. Make your own judgement.

#2 – The road to AI

From Operational to BI

We’ve had databases in our products for many years now. We use them to store data, run transactions and take actions. These are known as operational databases. For many years we’ve had another set of databases – the analytical ones, used in data warehouses. The reason we needed them is because they worked better when asking questions requiring aggregations that look at large series of historical data.

That got the marketing terms of BI (Business Intelligence) and even Analytics.

BI because we’re selling now to the business (at a higher price point of course). And what we’re selling is value.

Analytics because it sounds harder than the operational stuff.

From BI to BigData

The next leg of that journey started about a decade ago with BigData.

Storage started costing close to nothing, so it made sense to store everything. But now data warehouses from the good-ol’ BI days got too expensive and limiting. So we came out with BigData. Things like Hadoop and Cassandra came to be and we were happy again.

Now we could just throw all our data into Hadoop and run batch processes on it called MapReduce that ended up replacing/augmenting our data warehouses.

BigData was in big hype for some time. While it is very much alive today, it seems to have run out of steam for marketers. They moved on to Machine Learning.

From BigData to ML

This step is a bit more nuanced, and maybe it isn’t a step at all.

Machine Learning covers the research area of getting machines to decide on their own algorithm – or more accurately – decide on how an algorithm will be used based on a given dataset.

Machine learning algorithms have been around well before machines. If you check the notes on Wikipedia for Linear Regression, you’ll find the earliest methods for it were published in 1805. And to be fair, these algorithms are used in BI as well.

The leap from BigData to ML happened mostly because of Deep Learning. Which I am keeping as a separate leap from ML. Why? Because many of the things we do today end up being simpler ML algorithms. We just call it AI (or ML) just because.

Deep Learning got everyone on the ML bandwagon.

From ML to Deep Learning

Deep Learning is a branch of Machine Learning. A certain type of machine learning algorithms.

They became widely popular in recent years since they enabled the accuracy of certain tasks to increase significantly.

There are two things we can now achieve due to deep learning:

1. Better image classification
2. Better accuracy in speech to text

Here’s how Google fairs now (taken from KPCB internet trends):

We’ve been around the 70% accuracy at 2010, after a gradual rise in the past 40 years or so from 50%.

This steep rise in accuracy in this decade is attributed to the wide use of machine learning and the amount of data available as training material to the algorithms.

Deep learning is usually explained as neural networks, making it akin to human thinking (at least until the next wave of better algorithms will be invented which are more akin to human thinking).

From Deep Learning to AI

And then there’s artificial intelligence.

Less a specific algorithm and more a target. To replace humans. Or to do what humans can do.

Or my favorite:

AI is a definition of what we can’t do with machines today.

Once we figure that out, we’ll just put AI on the next pedestal so we’ll have a target to conquer.

#3 – Learning or Imitating?

Here’s one that is slightly different. I heard it at a data science event a couple of weeks ago.

Machine Learning is about getting machines to select their own algorithm by presenting them a set of rules and outcomes:

• You give a machine voice recordings, along with the transcription. And let them decide from that input on a new voice recording what the transcription should be
• You give a machine the rules to play a game, and let it play many times (millions?) until he gets better at it, devising his own algorithm and strategy

Artificial Intelligence is about doing something a human can do. Probably with the intent to replace him by automating the specific task. Think about autonomous driving – we’re not changing the roads or the rules of driving, we just want a car to drive itself the way a human would (we actually want the machine to drive better than humans).

So:

• Machine Learning is about letting a machine devise his own algorithm based on data we give it
• Artificial Intelligence is about doing a task the way a human would

#4 – Predictions vs Actions

This one I saw at a recent event, which got me on this track of ML vs AI in the first place.

Machine Learning is about Predictions, while Artificial Intelligence is about Actions.

You can use machine learning to understand things, to classify them, predict and estimate. But once the time comes to act upon it, we’re in the realm of artificial intelligence.

It also indicates that any AI system needs ML to operate.

I am sure you can poke holes in this one, but it is useful in many ways.

Why do we care?

While I am not a stickler to such details, words do have meaning. It becomes an issue where everyone everywhere is doing AI but some end up with a Google Duplex while others show a rolling average on a single metric value.

If you are using communications and jumpstarting an AI initiative, then be sure to check out our upcoming report: AI in RTC.

Want to help us with our research AND get a free ebook AND have a chance to win one of five $100 Amazon gift cards?

Fill out our AI in RTC survey

The post ML vs AI: What’s the difference between machine learning and artificial intelligence? appeared first on BlogGeek.me.

Kamailio SIP Server v5.0.7 stable is out – a minor release including fixes in code and documentation since v5.0.6. The configuration file and database schema compatibility is preserved, which means you don’t have to change anything to update.Kamailio v5.0.7 is based on the latest version of GIT branch 5.0. We recommend those running previous 5.0.x or older versions to upgrade. There is no change that has to be done to configuration file or database structure comparing with the previous release of the v5.0 branch.Resources for Kamailio version 5.0.7Source tarballs are available at:

• https://www.kamailio.org/pub/kamailio/5.0.7/src/

Detailed changelog:

• https://www.kamailio.org/pub/kamailio/5.0.7/ChangeLog

Download via GIT: # git clone https://github.com/kamailio/kamailio kamailio
# cd kamailio
# git checkout -b 5.0 origin/5.0Relevant notes, binaries and packages will be uploaded at:

• https://www.kamailio.org/pub/kamailio/5.0.7/

Modules’ documentation:

• https://www.kamailio.org/docs/modules/5.0.x/

What is new in 5.0.x release series is summarized in the announcement of v5.0.0:

• https://www.kamailio.org/w/kamailio-v5-0-0-release-notes/

Note: the branch 5.0 is the previous stable branch. The latest stable branch is 5.1, at this time with v5.1.4 being released out of it. Be aware that you may need to change the configuration files and database structures from 5.0.x to 5.1.x. See more details about it at:

• https://www.kamailio.org/w/kamailio-v5-1-0-release-notes/

Thanks for flying Kamailio!

An interview with Alan Masarek, CEO of Vonage.

Doing these video interviews is fun, so when the opportunity arose to be at the Vonage headquarters in Holmdel, New Jersey, it made sense to ask for a video interview with Alan Masarek, the CEO of Vonage.

In this interview, I wanted to get Alan’s viewpoint about the space he is operating in, especially now, some two years after the acquisition of Nexmo. It is quite common to find UCaaS vendors then are heading towards the contact center. Many will even add APIs on top. Vonage is the only one who decided to acquire a dominant CPaaS vendor (Nexmo).

As usual, you’ll find the transcript right below the video.

I enjoyed the interview and the hospitality. I’d like to thank Alan and the team at Vonage for setting this one up.

Transcript

Tsahi: Hi. So I have got here today, Alan Masarek, CEO of Vonage at the Holmdel, Vonage Technology Center.

Alan: That’s correct. We’re thrilled to be here at our Vonage Technology Center. It’s a pleasure to be with you, Tsahi. Thank you.

Tsahi: Thank you for having me here. I have a question before we start and this really bugged me a bit during the time that I’ve learnt about you and about the company: You came from Google to Vonage.

Alan: Yup.

Tsahi: Why?

Alan: Well, first of all, if that’s the only thing that’s bugged you, that would be exceptional. But in all seriousness, what excited me when I was presented this opportunity when I was at Google … And I’d gotten to Google from selling my earlier company to them back in 2012. So I was a director in the Chrome and apps group and I was very involved in the whole rollout of what is now today, G Suite. We used to call it Google for Enterprise.

What intrigued me about coming here was the opportunity to take this almost iconic consumer brand company that built this amazing level of awareness around providing residential phone service and how you could take the brand and the network asset as well as the cash flow from consumer candidly, and use that to pivot into business. I always look at markets the same way. You sort of sit back and you say, “Is that market worth winning and do you have the assets to give you an ability to win it?”

So when you look at the broader business communications market, it’s a massive TAM growing very quickly. And then even when you look at the competitive set, I found the big companies in this set were pretty unfocused. Most of the competitors were smaller companies, had less brand awareness, less sort of national scope, less profitability. So you have this huge TAM, a surmountable competitive set, then you have these assets from consumer that we felt we could bring to bear to win and that’s exactly what we’ve been executing on, that’s what we saw when I was at Google, that’s what I came here to do.

Tsahi: So you’re actually staying in this area between consumer and enterprise. You did that at Google with acquisition and now here at Vonage, moving from consumer to businesses.

Alan: That’s correct. So the company that I sold to Google focused really in the prosumer and enterprise segment. So we were a productivity solution that individuals would use and corporations would use. Here, we obviously have moved very specifically from our roots in consumer, in residential, focused in business. When we began that pivot, we started with small companies because that’s where the action was and the move to cloud, but now we’ve moved very purposefully upmarket to larger and larger corporate customers.

Last year, we signed what I think is the largest deal ever done in cloud communications with the largest residential real estate company in the United States. 21,000 corporate seats moving from prem to cloud and another 125,000 franchise seats.

Tsahi: Interesting. And what gets you up in the morning?

Alan: Well, this morning at 5 o’clock, my alarm clock but … What I’m excited about and I’ve continued … The reason I came here to begin with is I want to build a remarkable company here. It’s not just the transformation from moving from a residential-focused company to a business-focused company. We’re clearly executing on all those elements, whether it’s the technology platform itself, sales execution, the post-sales experience we provide our customers, all those things that we’re doing. But as important and in some respects if not more important, it’s the cultural transformation as well.

What I find that is really sort of stimulating to me is to create that switched-on Silicon Valley mindset culture. I like to think that we’re a billion dollar startup is what we talk about it. Last year, we finally crossed the billion dollar in revenue threshold. But I want to have the agility, the speed, the openness, the transparency, the honesty, all that, in order for Vonage to be … The way I describe it is I want Vonage to be that destination place to work the way Google was and everybody celebrates when they get a Google. I want them to feel the same way getting a job here.

Tsahi: Okay. And you’re a cloud communication company at the end of the day and cloud communication in the last few years have got a lot of attention, especially this last year. How come most of the businesses today are still on-premise when it comes to their communication needs?

Alan: On the communication side, the move to cloud has happened more slowly than CRM and ERP and HRM software, things like that. I think because the nature of dial tone has been about as reliable as the sun coming up tomorrow and there’s a great degree of risk that’s associated with it. Companies sit back and they say, “My goodness. It works. I don’t necessarily want to change it.” Now, the reality is when you move from the traditional prem-based solutions and the old PSTN network and such to IP-based, cloud-based solutions, you have infinite scalability, much, much more functionality, the whole notion of unified communications and communications platform as a service all stems from that. But I just think there’s been a fear factor that has caused it to migrate to the cloud more slowly than some of these other verticals.

But you see this amazing tipping point as recently as five years ago, only small companies for the most part were moving to the cloud. Now it has moved all the way up to major enterprises. And there are just example after example of other huge companies, global multinationals moving to cloud. It’s sort of no longer in dispute that cloud will supplant prem. It’s just like anything takes time.

Tsahi: What triggers them to do that shift, that migration from on-prem to cloud?

Alan: There are several trigger points. A couple of them are the comfort of moving to cloud. The cloud was scary just a few years ago and so it was to be avoided by bigger companies. But beyond that, it’s the productivity that they can get. Every company out there is going through their own digital transformation of one form or the other. Everybody is looking over their shoulder, scared to death of the more digitally transformed competitor has a bullseye on their back, is coming after their business. Obviously, we can always cite the example of physical retail stores versus Amazon eCommerce. That notion of digital transformation everyone has to go through and I think what’s happened is up until very recently, communications has been sort of the underappreciated element of digital transformation.

I always have this sort of visual metaphor in my mind that you can picture somebody on the old black rotary dial phone talking to a colleague saying, “We got to get that eCommerce site up.” Not realizing that the problem itself or a major piece of the problem itself is their communications infrastructure, how people work differently with one another, how they collaborate, et cetera, et cetera. All those elements of what we’re providing with these cloud communications solutions are fueling their digital transformations. I think that’s now being seen. Folks are more aware of that all the time and that’s why you’re seeing kind of everything change and move to cloud so quickly.

Tsahi: When you look at the communication market, for me, it’s like a Venn diagram with different parts of it. There’re unified communication and then contact centers and recently, we see APIs, these CPaaS communication platform as a service. When I look at what competitors do in this space, your competitors and unified communications, they end up going and doing something or adding stuff in the contact center. And then when they look at the APIs, usually go and say, “Well, we just put an API”; obviously they do because 2018, everybody uses an API on top of what they do. But you did something differently. You went and acquired the company called Nexmo and then their APIs, haven’t even touched it in a way and you left that to be a separate part of the business or a business all its own, with and without relationship to what you’re doing in unified communications.

Alan: The reason that we bought Nexmo is we have a view of what business communications is and will be that’s different than most. Most in the example have hosted PBX which has really been the principal use case of UCaaS or hosted contact center which has been the principal use case of CCaaS. In our view, those are just applications. Hosted PBX, moving your prem-based PBX to the cloud is a big TAM onto itself but it’s not necessarily an industry. The same applies to contact center. It’s not an industry. It’s simply an application or a use case which is really large and really important. But at the same token, the whole now new acronym of CPaaS, Communications Platform as a Service, says, “Well, there are other elements of communications that I want to simply program into my workflow, my mobile app, my business process, my website.” What have you. But have nothing to do with the contact center or the PBX.

Our view has been that we’re building a communications platform company. The whole notion of it is it’s a microservices architected platform. So we’re taking the Nexmo platform and our own Vonage Business Cloud platform and bringing those together. We refer to that internally as 1V, One Vonage. From that microservices architecture, you’re just going to serve customers in those big use cases. So whether you bundle several hundred of those microservices together in a use case called PBX or in a use case called contact center, or sell them one at a time that just get embedded into something else via the software APIs, it doesn’t matter. It’s the same platform. You’re just feeding where the needs are the greatest.

And the notion of this is that there’s not different industries, UCaaS, CCaaS, CPaaS. It’s simply communication elements, how they get deployed. The way I like to think about it is I go back to the music industry. We grew up, here’s songs and we can buy it only one way. Packaged, pre-published on an album. Apple came along and the cloud and said, “I’m going to unbundle the model and you can buy a song one at a time.” And then streaming services and subscription services have come along and the ability to mash up your music. They’re just different delivery models of the same song. It’s the way I think about cloud communications. There are communication elements, audio, video, messaging. Whether you package them in big applications like PBX or unbundle them as microservices, which is the CPaaS model, it doesn’t really matter. It’s just where the needs are the greatest.

Because at the end of the day, communication only serves a purpose. Does it make the company more productive? Does it connect my customers in a more personalized way with me as a company? And does it drive better business outcomes for my business? If it doesn’t do that, it doesn’t really matter whether you call it UCaaS or CCaaS or CPaaS. It simply has to drive those better business outcomes and that’s the approach that we’re taking.

Tsahi: Talking about Nexmo, they are now 12, 18 months part of Vonage now.

Alan: Almost two years. June 5th will be two years.

Tsahi: What synergies have you seen since the acquisition, up until today?

Alan: There’s been a great deal of synergies. You mentioned before about the Venn diagrams where much of the industry has developed as if the segments, UCaaS, CCaaS, CPaaS have been separate. We reject that. If they were all Venn diagrams, they all will be separate. Our view is they’re coming together all the time. So increasingly, the purchaser at a company, Acme  company, is the line of business manager. The conventional wisdom used to be that if I’m buying UCaaS, I’m the CIO or the head of IT and if I’m buying CCaaS contact center, I’m the help center. And if I’m buying communications platform as a service, I’m an individual developer, perhaps even the CMO. What you’re finding now is it’s coming together as lines of business. Given that trend from a synergy point of view, we’ve organized since the acquisition, completely functionally so that the entire engineering team, Vonage traditional or Nexmo reports up to the same CTO. The product organization up to the same chief product officer. Sales under the same chief revenue officer, same with marketing.

And they’re already doing tremendous amounts of lead sharing within the groups, operational sharing, sales enablement, sales training and things like that. Because what we’re finding is that in the cloud PBX world, your salespeople don’t want to go out there and go to a customer and say, “Buy me because my hunt group or my auto attendant is better than the other guys.” Because this very sort of baseline functionality. What you want to do is go into your customer and have a conversation about better business outcomes. So they’re just naturally carrying Nexmo into the discussion with every prospect out there. You can look at every one of our large company wins. It began with a Nexmo conversation interestingly, more than just the feature set of the PBX or the contact center. So you’re seeing very, very natural synergies happen. Now, it’s not a cost synergy issue for us in terms of people. When we bought Nexmo, it was about 175 people. I think it’s above 300 today and as I recall last time when I was in our London office, there was 140 open jobs for Nexmo this calendar year, so we’re growing in a big hurry.

Tsahi: We’ve talked about the cloud, we’ve talked about API. There is another big buzzword these days around communications and that’s “Teams”. The notion of what Slack started in a way. Messaging inside groups, smaller groups which is more ad hoc than the usual grounded structured way of communications. And you see today Microsoft going there, Cisco going there. All the big companies are headed there and then next to you, you got Google and Amazon joining this specific space. How is Vonage preparing towards that future of team collaboration, enterprise messaging, whatever you want to call it?

Alan: So not to sort of disclose all the goodies that are coming but within our roadmap, we have some very, very interesting developments around the collaboration and work stream messaging space that will be coming out later this year. And that’s tightly integrated as a single app whether you’re mobile, desktop or browser, with the experience in the communications system. Now, it also will integrate well with the major players that you just talked about. Slack, Stride, Teams, et cetera. Or it’s going to be WebEx, et cetera. Because it has to.

In our view, we can’t play king maker and say, “Oh. Mr. Customer, Mrs. Customer, you cannot use these other collaboration tools.” That’s ultimately going to the decision of the customer. So we have to have our own solution that is built-in in a fully integrated way but then the ability to integrate in with the others and that’s the approach that we’re taking.

Tsahi: Can I ask a question that just occurred to me?

Alan: Sure.

Tsahi: What about contact centers?

Alan: I think contact center is incredibly important as part of the integrated solution. And so today, we have a contact center built into Vonage Business Cloud which is our own proprietary call processing stack. And for our Vonage Enterprise Solution, we use BroadWorks contact center functionality. Then, in those situations where they need an advanced contact center solution, then we are a reseller of inContact. But again, it’s integrated fully in with our solution, so it appears like it’s a single experience. And then we serve it as if it’s a single experience so the contract is on our paper, the support is ours, things like that.

Contact center though becomes very, very important in the CPaaS market because so much of how communications get embedded in through some software API into that website, that mobile app, business process, what have you, is about customer experience. And so think of it as task routing. Somebody is on my website and they’re looking at my product and they have a question. Today, they may pick up the phone and call and have to start over because there was no context to what they were doing on the website, and these CPaaS type tools are all about the contextual. The software identifies the context to what I was doing.
So if was on Delta Airlines site trying to book a flight and I was 10 minutes into booking the itinerary and all of a sudden it had a problem, in the past, I’d pick up the phone and just call and have to start over because no one had any idea of the itinerary I was just trying to book. These new contextual tools that you can embed in, understand the itinerary so that it routes through the appropriate IVR into the contact center. So think of it as a task, an intelligent task. It knows I was trying to book a flight from Tokyo to Shanghai next Thursday and it will route me through the appropriate IVR to the person on the help desk for the international Asia markets.

And so you can envision from a customer personalization or a customer intimacy, rather than me having to start over which is what happens today, which is very frustrating to all of us. You can imagine the agent picking the phone up and saying, “Hi, Mr. Masarek. I see you’re trying to book a flight next Thursday from Tokyo to Shanghai. How can I help?” That’s a direct connection between the customer experience, routing the task into the contact center. We think that’s very important.

Tsahi: Let’s look a little bit into the future.

Alan: Okay.

Tsahi: What do you think is the biggest challenge for the modern businesses moving forward from now on? When it comes to communications of course.

Alan: I’m not sure it’s a challenge. I don’t want to sort of split words between challenge and opportunity, but I actually think communications is going to fundamentally change by virtue of we’re no longer tethered to a physical device. We think about communications, I’m on a call, either a landline or a desk. In our vision for it, communications is in everything. So whether it’s a click-to-call or click-to-communicate functionality in the website or … Pick whatever app you want. You’re on Salesforce, I’m on an Excel spreadsheet, someone else is in G Suite or in Gmail, or in Google Sheets. Doesn’t matter. There will be click-to-communicate functionality everywhere and naturally, these microservices that are going to be created increasingly by these CPaaS type solutions. So you’re going to have I think this explosion in communications the way I think about it because you’re no longer tethered to anything physical. You’re in an app or a website or what have you.

And the way I think about it is your decision of how you communicate is simply going to be a function of the limitations of the physical device that you got onto the internet with. So for instance, if the device doesn’t have a camera, you’re not going to do video. If it doesn’t have a speaker and microphone, you’re only going to do messaging, that’s all you can. But the mode, video, audio or messaging is going to be the limitations of the device and your personal preference, also kind of situational. If you just stepped out of the shower, you’re not going to do video likely. So the point is regardless of how you’re interacting in some sort of app or website, you’re going have communication everywhere. So I think the notion of the challenge to companies is less the challenge and more that I think it’s going to change the way we work because the notion of how we collaborate, how we share, the tightness of the communication, sort of that feedback loop is going to get tighter, and tighter, and tighter is the way I think about it.

I actually think about communication, this renaissance or this explosion in communication a little bit like the internet 10 years ago. 10 years ago, there was no video flying around the internet. It was kind of more flat files and such. There wasn’t full-motion video. There certainly wasn’t virtual reality and things like that, and self-driving cars and all these stuff that is just massive quantities of data that are going around the internet. When that began, look what happened with all the content delivery networks. They just kind of went like this in terms of the volume of capacity they have on the internet. I think communications is going to go through this similar renaissance or explosion in the sense because if communications are everywhere, not just on specific devices, you’re going to be communicating all the time, and so I think you’re going to see this massive uplift in it. If it’s a challenge out there, it’s going to create sort of communication overload, perhaps, but maybe smarter people than use will figure it out on how to make it simpler.

Tsahi: And moving forward, would businesses end up building their communication needs on top of APIs, go pick a UCaaS or a communication solution to do that for them or go for even a very specific niche SaaS product to get what they need?

Alan: I think that increasingly, communications will be built on top of the platform, the PaaS product, not going and buying some monolithic application. Like you said earlier, everybody’s got APIs. The old way we used to write software, we write a big monolithic solution from the UI, the user interface, all the way down to the metal called PBX, in our example. I can open up APIs to the PBX but it’s not programmable. It’s simply an API into that monolithic solution. Where we sit today is a microservices architecture where it’s fully programmable.

And I think what you’ll see, and this is exactly the strategy we’re building to, is whether you want to use that big chunk of microservices in a particular use case that is as a big application like PBX or a big application like contact center, it’s just a function of what’s the best way to deliver it to a customer. Do I think people are going to build their own PBX all the time? No. Because I think to me it’s analogous to the vast majority of people don’t build their own computer. You certainly could. You could be a hobbyist and build your own PC and buy the motherboard and the chassis and the whole bit, but very few people do that when you go out and buy a computer for $400. So I think the PBX distribution model where it’s something you’re going to subscribe to, it’s a SaaS solution, will persist, but I think the microservices are really going to takeover where communications get woven into everything else.

Tsahi: Vonage in 5 to 10 years from now, where do you see the company itself? What are you going to sell to businesses, to consumers? What kind of services are going to be there?

Alan: Vonage in the next five years will be an extraordinarily different company than it is today. Let me go backwards first. Four years ago, we were 100% consumer. Now, this year in 2018, roughly 60% of the revenue is business. Business is growing really quickly. So as of last quarter, 22% growth organically, nothing to do with acquisitions. And consumer has been declining as residential home phone usage is in decline, by 12% roughly. Now that business is the larger of the two segments and growing at twice the rate that consumer’s declining, you can imagine where the line separate in a very big hurry. So the whole focus of the organization is on business. It already is. Consumer is still a meaningful piece, it’s 40% but it’s getting smaller all the time as a percentage of the total.

What’s interesting from a how we’re going to serve customers is precisely the way we do it today. Our whole approach from a platform perspective, the way I described it where irrespective of whether it’s UCaaS, CCaaS or CPaaS, coming out of a common platform, we will continue to execute on that. What’s interesting where I think a value unlock happens for the company is you’re now going to have … We’re already having consolidated revenue growth.

Last year, we did just above a billion dollars in revenue. This year, Wall Street has us close to a billion fifty. Again, as the smaller piece, consumer, get smaller and smaller, it’s mitigating impact and overall growth declines. Therefore, we’re sort of more and more of a consolidated growth company. Again, unrelated to any acquisitions, just purely organically. The notion then of, “Oh my goodness. You’re in the midst of a transformation” goes away because you’ve now transformed.

So where I can see us in pretty short order is serving our approach to our customers in this differentiated way which I think will withstand the test of time, will withstand competitive entrance because, the end of the day, we’re just rooted in how do we provide better business outcomes for our customers. But now you’re going to have this increasingly fast growing consolidated company, well greater than a billion dollars in revenue, highly profitable still and I think that’s going to be a value unlock for the story. When I go back to many transformational stories in the early days, there’s a lot of investor skepticism about transformational stories is most of them don’t work. This one’s worked and that’s why we’ve had sort of a almost quadrupling of our stock price over the last four years.

Tsahi: Okay.

Alan: All right.

Tsahi: Thanks for your time, Alan.

Alan: My pleasure. Thanks so much. I enjoyed it.

Tsahi: Me too.

Alan: Sure. Thank you.

Tsahi: Thank you.

The post UCaaS, CCaaS & CPaaS: An interview with Alan Masarek, Vonage CEO appeared first on BlogGeek.me.

To discuss about the next major release of Kamailio (v5.2.0) and ongoing development, we propose an IRC devel meeting for next week, on Tuesday, Jun 26, 2018.The meeting is going to be held as usual in the #kamailio channel on freenode.net IRC network.A wiki page was created to collect the topics that are wanted to be discussed:

• https://www.kamailio.org/wiki/devel/irc-meetings/2018a

Fell free to add there or reply to the mailing list with what you think it is relevant to discuss.Thanks for flying Kamailio!

The Chrome Webstore has decided to stop allowing inline installation for Chrome extensions. This has quite an impact on WebRTC applications since screensharing in Chrome currently requires an extension. Will the [crayon-5b2272a8d9b0f447286991-i/] API come to the rescue? Screensharing in Chrome When screensharing was introduced in Chrome 33, it required implementation via an extension as a way to […]

The post Chrome Screensharing Blues – preparing for getDisplayMedia appeared first on webrtcHacks.

Now that it is getting relatively easy to setup video calls (most of the time), we can move on to doing fun things with the video stream. With new advancements in Machine Learning (ML) and a growing number of API’s and libraries out there, computer vision is also getting  easier to do. Google’s ML Kit is […]

The post Smile, You’re on WebRTC – Using ML Kit for Smile Detection appeared first on webrtcHacks.

ML in RTC can fit anywhere – from low level optimization to the higher application layers.

TL;DR – I am working with Chad Hart on a new ML in RTC report. If you are interested in it, scroll down to the end of this article.

Machine Learning (ML), Artificial Intelligence (AI), Big Data Analytics. Call it what you will. You’ll be finding it everywhere. Autonomous cars, ecommerce websites, healthcare – the list goes on. In recent years we’ve seen a flourish in this domain due to the increase in memory and processing power, but also due to some interesting breakthrough in machine learning algorithms – breakthroughs that have rapidly increased the accuracy of what a machine can now do.

My ML Origin Story

I’ve been looking and dealing with machine learning for many years now. Never directly calling it that, but always in the vicinity of the communications industry.

It probably started in university. I decided to do an M.Sc because I was somewhat bored at work. I took a course in computational linguistics which then ended with me doing research in backward transliteration, looking at phonemic similarities between English and Spanish (#truestory). That was in 2005, and we used a variant of dynamic programming and the viterbi algorithm. That and other topics such as hidden markov model were my part and parcel at the time.

Later on, I researched the domain of Big Data and Analytics at Amdocs. I was part of a larger group trying to understand what these mean in telecommunications. Since then, that effort grew into a full business group within Amdocs (as well as the acquisition of Pontis, well after I left Amdocs for independent consulting).

Which is why when I talked to Chad Hart about what we can do together, we came to an agreement that something around ML and AI made a lot of sense for both of us, and taking it through the prism of RTC (real time communications), placed it in the comfort zone of both of us.

We molded that effort under the Kranky Geek roof, calling it Kranky Geek Research. Created a landing page for our research, a brochure and a survey (more on that later).

During that period, we thought a lot about what domains we wish to cover and what ML in RTC really means.

Categorizing ML in RTC

Communications is a broad enough topic, even when limited to the type that involves humans. So we limited even further to real time communications – RTC. And while at it, threw text out the window (or at the very least decided that it must include voice and video).

Why do that? So we don’t have to deal with the chatbots craze. That’s too broad of a topic on its own, and we figured there should be quite a few reports there already – and a few oil snake sellers as well. Not our cup of tea.

This still left the interesting question – what exactly can you do with AI and ML in RTC?

We set out to look at the various vendors out there and understand what are they doing when it comes to ML in RTC.

Our decision was to model it around 4 domains: Speech Analytics, Computer Vision, Voice Bots / Assistants and RTC quality / cost optimization.

1. Speech Analytics

Speech Analytics deals a lot with Natural Language Processing (NLP) and Natural Language Understanding (NLU).

Each has a ton of different use cases and algorithms to it.

Think of a contact center and what you can do there with speech analytics:

• Employ speech-to-text for transcription of the sessions
• Go further with sentiment analysis from analyzing voice queues and not only the transcripted text
• Glean meaning out of the transcription and glean actionable insights based on that meaning

You will find a lot of speech analytics related RTC ML taking place in contact centers. A bit less of it in unified communications, though that might be changing if you factor in Dialpad’s acquisition of TalkIQ.

2. Computer Vision

Computer Vision deals a lot with object classification and face detection, with all the derivative use cases you can bring to bear from it.

“Simple” things like face recognition or emotion recognition can be used in real time communications for a multitude of communication applications. Object detection and classification can be used in augmented reality scenarios, where you want to mark and emphasize certain elements in the scene.

Compared to speech analytics, computer vision is still nascent, though moving rapidly forward. You’ll find a growing number of startups in this domain as well as the cloud platform giants.

3. Voice Bots & Assistants

To me, voice bots and assistants is the tier that comes right above speech analytics.

If speech analytics gets you to NLP and NLU, the ability to convert speech to text and from there moving to intent. Voice bots are about conversations – moving from a single request to a fluid interaction. The best example? Probably the Google Duplex demo – the future of what conversational AI may feel like.

Voice bots and assistants are rather new to the scene and they bring with them another challenge – do you build them as a closed application or do you latch on to the new voice bot ecosystems that have been rapidly making headway? How do you factor in the likes of Amazon Alexa, Google Home, Google Assistant, Siri and Cortana into your planning? Are they going to be the interaction points of your customers? Does building your own independent voice bot even makes sense?

Whatever the answers are, I am pretty sure there’s a voice bot in the future of your communications application. Maybe not in 2018, but down the road this is something you’ll need to plan for.

4. RTC Quality & Cost Optimizations

While the previous 3 machine learning domain areas revolve around new use cases, scenarios and applications through enabling technologies, this one is all about optimization.

There are many areas in real time communication that are built around heuristics or simple rule engines. To give an example, when we compress and decompress media we do so using a codec. The encoding process (=compression) is lossy in nature. We don’t keep all the data from the original media, but rather throw away stuff we assume won’t be noticed anyway (sounds outside the human hearing range, small changes in color tones, etc) and then we compress the data.

The codecs we use for that purpose are defined by the decoder – by what you do if you receive a compressed bitstream. No one is defining when an encoder needs to look like or behave. That is left to developers to decide, and ecoders differ in many ways. They can’t brute-force their way to the best possible media quality, especially not in real-time – there’s not enough time to do that. So they end up being built around guesswork and heuristics.

Can we improve this with machine learning? Definitely.

Can we improve network routing, bandwidth estimation, echo cancellation and the myriad of other algorithms necessary in real time communications using machine learning? Sure we can.

The result is that you get better media quality and user experience by optimizing under the hood. Not many do it, as the work isn’t as high profile as the other domains. That said, it is necessary.

Interested in ML in RTC?

Here are a few things you can do:

Fill out our survey

This will get factored into the quantitative part of our report. If you fill it out, you will also receive a complimentary e-book we’re writing titled Intro to AI in RTC.

Take the ML in RTC Survey

Learn more about the report

Interested in the report itself? Thinking of purchasing it? Great! We have a special launch discount.

You can find more information about the report itself in our research page.

Download the report prospectus here

Share your opinion on AI in RTC

Doing something interesting in this space? Share your thoughts with us.

Contact us via research@krankygeek.com to participate in our study.

The post Where does Machine Learning fit in Real Time Communication (ML in RTC)? appeared first on BlogGeek.me.

Kamailio SIP Server v5.1.4 stable is out – a minor release including fixes in code and documentation since v5.1.3. The configuration file and database schema compatibility is preserved, which means you don’t have to change anything to update.Kamailio® v5.1.4 is based on the latest source code of GIT branch 5.1 and it represents the latest stable version. We recommend those running previous 5.1.x or older versions to upgrade. There is no change that has to be done to configuration file or database structure comparing with the previous releases of the v5.1 branch.Resources for Kamailio version 5.1.4Source tarballs are available at:

• https://www.kamailio.org/pub/kamailio/5.1.4/src/

Detailed changelog:

• https://www.kamailio.org/pub/kamailio/5.1.4/ChangeLog

Download via GIT: # git clone https://github.com/kamailio/kamailio kamailio
# cd kamailio
# git checkout -b 5.1 origin/5.1Relevant notes, binaries and packages will be uploaded at:

• https://www.kamailio.org/pub/kamailio/5.1.4/

Modules’ documentation:

• https://www.kamailio.org/docs/modules/5.1.x/

What is new in 5.1.x release series is summarized in the announcement of v5.1.0:

• https://www.kamailio.org/w/kamailio-v5-1-0-release-notes/

Thanks for flying Kamailio!

What should you be doing about the upcoming WebRTC 1.0 release?

That comic strip above? I think it embodies nicely what comes next.

We’ve started with WebRTC somewhere in 2011 or 2012. Depends who’s counting. So we’re 6 or 7 years in now.

I’ve been promised WebRTC 1.0 in 2015 I think.

Then again in 2016.

In 2017, I was told that WebRTC 1.0 is just around the corner. Definitely going to happen before year end.

Guess what? We’re now almost halfway through 2018. And no WebRTC 1.0. Yet.

But it is coming.

Look at this message on discuss-webrtc from April this year: Transitioning Native PeerConnection to WebRTC 1.0. Google are shifting from what is called Plan B to Unified Plan.

To give you the gist, Google will be ripping out some code, adding new code. Removing APIs. Modifying others. The timeline stated for all this in that posting?

• End of April 2018: “Unified Plan” and the new APIs stabilizes
• July 2018: Default SdpSemantics changes to UnifiedPlan
• No earlier than end of year 2018: PlanB semantics removed and UnifiedPlan becomes the only option

Change is in the air…

That change is going to affect developers and testers everywhere, and the end result is going to be uncertainties and surprises in the coming months. How many months? Many months

There’s not much you can do about it besides allocating resources to the problem in the short and mid term future. These resources should be in development and testing.

I touched development in a previous webinar I did with Philipp Hancke when I launched the next round of my WebRTC training.

Now I want to talk about preparation aspects in the testing domain – what exactly you should be expecting moving forward.

To that end, I am a visitor of the upcoming WebRTC Standards webinar series. The webinar takes place later today –

Register for free now

The post Preparing for the WebRTC 1.0 – a free webinar appeared first on BlogGeek.me.

A new module has been imported in Kamailio’s git repository, respectively app_ruby. It allows embedded execution of Ruby scripts, either inline inside a native kamailio.cfg or as a KEMI routing logic.The readme of the module is available at:

• https://www.kamailio.org/docs/modules/devel/modules/app_ruby.html

The idea of the module was suggested during Kamailio World Conference 2018, couple of attendees mentioning that they are mainly Ruby developers and KEMI framework didn’t support it. With some hints given by some of them, it proved that developing such module was not that complex. It is still in early stage, but testing can be done and feedback is very appreciated.The KEMI exported functions by the core and all the other modules are already available, some work needs to be done for KSR.x submodule, which is going to be exported by app_ruby itself.Now the list of scripting languages supported by the KEMI framework are: Lua, JavaScript, Python (v2 and v3), Ruby and Squirrel.Enjoy! Thanks for flying Kamailio!

Shortly after the end of the 6th edition of Kamailio World Conference (May 14-16, 2018, in Berlin, Germany), the video recordings of the sessions were published on KamailioWorld Channel on Youtube:

• https://www.youtube.com/channel/UCElq4JNTPd7bs2vbfAAYVJA/videos

The channel is a valuable knowledge base, only from Kamailio World 2018 there are over 35 recorded sessions, approaching topics such as IMS/VoLTE, emergency services, use of Kamailio with Redis, SIP routing scripting with KEMI (Lua, Python, JavaScript, …), least cost routing engine, deployments and continuous integration with Docker, integration with other systems such as Asterisk, FreeSwitch, CGRateS or SipCapture. All these add to the videos from the past Kamailio World editions.If you haven’t followed the channel, you should do it and watch the videos on topics of interest for you, there are many tips and hints shared by developers and experienced users of Kamailio.Enjoy!Thanks for flying Kamailio!

Building an interactive application? There’s more than one WebRTC programming language that can fit your needs.

Last time I’ve written about WebRTC programming languages it was some two years ago. My focus then was how programming languages fit to different WebRTC components. That article is still relevant, so I suggest you read it as well.

This time, I want to focus on something slightly different. In recent months I’ve had the pleasure of watching as well as consulting teams of developers who are using programming languages that don’t always make sense to me when it comes to WebRTC. And yet, once they explained their reasoning, the decision and path they took would be one I couldn’t just dismiss if I were in their shoes.

Instead of doing this one by component, lets do it by language:

JavaScript

When it comes to languages, JavaScript is probably the most popular one these days. Don’t just take my word for it – look at Stack Overflow’s annual survey:

JavaScript is a winner with a huge margin on any other technology. It is also what you need if you are planning to make use of WebRTC in the browser.

There are two main places where you see JaveScript in WebRTC apps today: client side as well as signaling server.

Clients

Client side is simple enough. It is due to the fact that this is how you use WebRTC in browsers, but it can also be seen in cross platform mobile development (though not that much) or when using Electron for WebRTC.

Signaling

Signaling servers is about Node.js. And yes, the guest article from 2013 (!) is as relevant now as it was then. Twelephone (mentioned in that article) is long dead, and Chris has moved on to IOT and later decentralized networks. But the use of Node.js as a signaling server for WebRTC is going strong. If you take that route, just make sure to pick a popular framework.

Media Servers

I’ve seen Node.js being used in media servers as well:

• mediasoup for example, is an open source SFU that was built as a Node.js module to fit into a larger application
• SwitchRTC, a commercial SFU that got acquired by YouNow, was a combination of C/C++ and Node.js
• appear.in’s SFU (available for their PRO accounts) was built using Node.js

C/C++

I’ll be using C and C++ together here, as I don’t see a huge distinction between the two (and in most cases, those that develop code in C create C++ abstractions, and those developing in C++ end up writing C code anyways).

C/C++ is a kind of a lowest common denominator. My guess is that a lot of the languages here have their compilers/interpreters written in C/C++ anyways. It is also a language that is available everywhere, though not always accessible directly.

Clients

Go to webrtc.org and you’ll find out that the code WebRTC that Google open sourced is written in C++. The easiest thing to do if you want to support WebRTC on an embedded device is probably to start by taking that code and hammering it until it fits your device.

It isn’t the only way to get WebRTC into a device but it sure is a popular one.

Signaling

Signaling with C++ isn’t common. Where you will find it is when SIP meets WebRTC.

Interestingly, all main open source SIP servers are written in C/C++: Asterisk, OpenSIPS and FreeSWITCH.

I am assuming this is because they are older than other WebRTC signaling implementations that tend to use higher languages.

STUN/TURN

NAT traversal servers are written today in C/C++.

Not all of them. But the most popular one is (coturn).

Media Servers

Media servers need to be highly performant, which is why most of them also end up being written in C/C++ – at least the parts that matter for performance.

• Janus, a popular media server is written in C
• SwitchRTC mentioned in relation to Node.js has a C++ component to it handling all media networking
• Kurento’s core is written in C/C++
• CPaaS vendors are using C/C++ in their media servers – at least those that I know the programming languages that they use

Java

Java is the most popular language based on the TIOBE index.

I don’t like Java that much, probably due to its verbosity and all the pain that its garbage collection causes to real time apps. But I don’t get to decide what others use

Java is probably one of the most popular programming languages also in WebRTC backend development, but not only there.

Clients

Android requires Java, which means Android native development with WebRTC also requires Java coding.

Besides this obvious option, there’s also the part of writing WebRTC clients in Java from scratch.

While you won’t find any open source Java client for WebRTC, I know of two separate WebRTC client implementations that use Java.

Signaling

There are many signaling servers out there that end up using Java when it comes to WebRTC. You’ll find it in enterprise software but not only.

A few of them are also open sourced, though there isn’t a specific one that is widely popular or highly recommended as far as I can tell.

Media Servers

Several of the media servers out there use Java. Either for everything or for the higher level abstractions.

Here are two of them:

• Jitsi is written in Java
• OpenVidu, an SFU implemented on top of Kurento by the Kurento maintainers is written in Java

Swift

If Java is needed for development of WebRTC in Android, then Swift is what you need for iOS.

Unless you’re fine with using Objective-C and not adopt Swift.

Other than iOS, I haven’t seen Swift used anywhere else when it comes to WebRTC implementations (or otherwise for that matter).

Python

As a higher level language, Python gets high marks. I’ve been introduced to it about a decade ago and loved it since, though I can’t say I used it too much myself.

When it comes to WebRTC, Python has its role in signaling servers, as many other higher level languages.

The most notable Python project related to WebRTC is [matrix]. Its open source Synapse server implementation is written in Python.

There are others using Python for their signaling. My guess is that this is just because of familiarity with the language.

Ruby

You can’t mention Python without mentioning Ruby.

I am guessing that Ruby can fit in signaling servers just as well as Python. Only difference is that I know of no one that is doing that.

C#

If someone is making use of a Windows based development stack, he is more likely to use C# than anything else.

In such cases, you will see the use of C# for all aspects of WebRTC – from native WebRTC client implementations, through signaling, to NAT traversal and media servers.

I am not a big fan of backend development on Windows, but if you are one of those who need it, then know you are not alone.

PHP

Yes. PHP.

PHP has gotten a bad rep when it comes to programming languages. It sits well with the fact that over 80% of all the websites use PHP:

Having such huge market share means more haters, especially when the language itself isn’t the most modern one out there.

What surprised me (it shouldn’t, but it did), was that there are companies who use PHP for their signaling server when it comes to WebRTC. I would never have thought that would be the case, but it is.

If you want to use PHP for your signaling server, then go for it. Just make sure you understand the limitations and implications of it.

Erlang

We’re getting into the more exotic alternatives.

Erlang is such a programming language to me. Created by Ericsson and open sourced ages ago, Erlang offers some interesting capabilities (go read Wikipedia).

There are a few projects out there that make use of Erlang for signaling and one for NAT traversal. I am not aware of anyone using Erlang in a production service, though I am sure there is.

Elixir

Elixir is built on top of Erlang (or at least on its virtual machine).

Only one I know who makes use of it for WebRTC is Slack.

In last year’s Kranky Geek event, Slack shared their plans of migrating from their Janus implementation to an in-house developed Elixir media server. You can watch that here:

Go

Go is a programming language created by Google. It is somewhere between C and C++ as much as I can tell (never been an expert in Go).

This one came to my attention with the recent implementation of STUN/TURN in Go – Pion TURN.

Not sure how popular is Go with WebRTC developers.

Where do these languages fit?

I tried taking the information above and placing it in an easy to use table, to give a quick summary of the state of WebRTC programming languages.

• Green indicates a popular choice
• Orange indicates an alternative that I’ve seen being used in the wild (or in production)
• Gray is something I haven’t seen used at all

Which language should you use?

I am sure I missed a language or two. And I am also sure some people are using WebRTC programing languages differently than I’ve described here. Feel free to share in the comments for this article or by emailing me about it – I’d love to learn more.

What programming language should you use?

That’s a different question. I’d say it depends on several factors:

• What is it you are implementing and for which devices?
• What do your current developers know and are comfortable with?
• What is your operational envelope for the service?
• Are there any popular open source or commercial products for WebRTC in that language?
• How easy will it be to find experienced developers for that language? How about developers who know both this language and WebRTC?

The post WebRTC Programming Languages and Where they fit appeared first on BlogGeek.me.

There will be a short maintenance timeframe for kamailio.org server schedule for later today, May 25, 2018, at 18:00 CEST (16:00 GMT). The website, including the wiki, as well as the mailing lists are going to be affected.The maintenance is expected to take no longer than 5 minutes.

Google in 2018 is all about AI. But not only…

In November 2015, Google released TensorFlow, an open source machine learning framework. While we’ve had machine learning before that – at Google and elsewhere, this probably marks the date when machine learning and as an extension AI got its current spurt of growth.

Some time, between that day and the recent Google I/O event, Sundar Pichai, CEO of Google, probably brought his management team, knocked on the table and told them: “We are now an AI company. I don’t care what it is that you are doing, come back next week and make sure you show me a roadmap of your product that has AI in it.”

I don’t know if that meeting happened in such a form or another, but I’d bet that’s what have been going at Google for over a year now, culminating at Google I/O 2018.

After the obligatory icebreaker about the burger emoji crisis, Pichai immediately went to the heart of the keynote – AI.

Google announced AI at last year’s Google I/O event, and it was time to show what came out of it a year later. Throughout the 106 minutes keynote, AI was mentioned time and time again.

That said, there was more to that Google I/O 2018 keynote than just AI.

Google touched at its keynote 3 main themes:

1. AI
2. Wellbeing
3. Fake news

I’d like to expand on each of these, as well as discuss parts of Smart Displays, Android P and Google Maps pieces of the keynote.

I’ll try in each section to highlight my own understanding and insights.

Before we begin

Many of the features announced are not released yet. Most of them will be available only closer to the end of the year.

Google’s goal was to show its AI power versus its competition more than anything else they wanted to share in this I/O event.

This is telling in a few ways:

1. Google weren’t ready with real product announcements for I/O that were interesting enough to fill 100 minutes of content. Or more accurately, they were more interested in showing off the upcoming AI stuff NOW and not wait for next year or release it later
2. Google either knows its competitors are aware of all the progress it is making, or doesn’t care if they know in advance. They are comfortable enough in their dominance in AI to announce work-in-progress as they feel the technology gap is wide enough

AI

When it comes to AI, Google is most probably the undisputed king today. Runners up include Amazon, Microsoft, IBM, Apple and Facebook (probably at that order, though I am not sure about that part).

If I try to put into a diagram the shift that is happening in the industry, it is probably this one:

Not many companies can claim AI. I’ll be using ML (Machine Learning) and AI (Artificial Intelligence) interchangeably throughout the rest of this article. I leave it to you to decide which of the two I mean

AI was featured in 5 different ways during the keynote:

1. Feature enhancer
2. Google Assistant (=voice/speech)
3. Google Lens (=vision)
4. HWaaS

Feature Enhancer

In each and every single thing that Google does today, there’s an attention to how AI can improve that thing that needs doing. During the keynote, AI related features in GMail, Google Photos and Android were announced.

It started off with four warm-up feel-good type use cases that weren’t exactly product announcements, but were setting the stage on how positive this AI theme is:

• Diagnosing diseases by analyzing human retina images in healthcare
• Predicting probability of rehospitalization of a patient in the next 24 hours
• Producing speaker based transcription by “watching” a video’s content
• Predictive morse typing for accessibility

From here on, most sections of the keynote had an AI theme to them.

Moving forward, product managers should think hard and long about what AI related capabilities and requirements do they need to add to the features of their products.What are you adding to your product that is making it SMARTER?

Google Assistant (=voice and speech)

Google Assistant took center stage at I/O 2018. This is how Google shines and differentiates itself from its main 3 competitors: Apple, Amazon and Facebook.

In March, Forbes broke some interesting news: at the time, Amazon was hiring more developers for Alexa than Google was hiring altogether. Alexa is Amazon’s successful voice assistant. And while Google hasn’t talked about Google Home, its main competitor at all, it did emphasize its technology differentiation. This emphasis at I/O was important not only for Google’s customers but also for its potential future workforce. AI developers are super hard to come by these days. Expertise is scarce and competition between companies on talent is fierce. Google needs to make itself attractive for such developers, and showing it is ahead of competition helps greatly here.

Google Assistant got some major upgrades this time around:

1. WaveNet. Google now offers an improved text to speech engine that makes its speech generator feel more natural. This means:
   1. To get new “voices” now requires Google to have less samples of a person speaking
   2. Which allowed it to introduce 6 new voices to its Assistant (at a lower effort and cost)
   3. To make a point of it, they started working with John Legend to get his voice to Assistant – his time is more expensive, and his voice “brand” is important to him, so letting Google use it shows his endorsement to Google’s text-to-speech technology
   4. This is the first step towards the ability to mimic the user’s own voice. More on that later, when I get to Google Duplex
2. Additional languages and countries. Google promised support for 30 languages and 80 countries for Assistant by year end
3. Naturally Conversational. Google’s speech to text engine now understand subtleties in conversations based not only on what is said but also how it is said, taking into account pitch, pace and pauses when people speak to it
4. Continued conversation. “Hey Google”. I don’t need to say these action words anymore when engaging in a back and forth conversation with you. And you maintain context between the questions I ask
5. Multiple actions. You can now ask the assistant to do multiple things at once. The assistant will now parse them properly

Besides these additions, where each can be seen as a huge step forward on its own right, Google came out with a demo of Google Duplex, something that is best explained with an audio recording straight from the keynote:

If you haven’t watched anything from the keynote, be sure to watch this short 4 minutes video clip.

There are a few things here that are interesting:

• This isn’t a general purpose “chatbot”/AI. It won’t pass a turing test. It won’t do anything but handling appointments
• And yet. It is better than anything we’ve seen before in doing this specific task
• It does that so naturally, that people can’t distinguish it from a real person, at least not easily
• It is also only a demo. There’s no release date to it. It stays in the domain of “we’ve got the best AI and we’re so sure of it that we don’t care of telling our competitors about it”
• People were interested in the ethical parts of it, which caused Google to backtrack somewhat later and indicate Duplex will announce itself as such at the beginning of an interaction
  • Since we’re still in concept stage, I don’t see the problem
  • I wouldn’t say google were unethical – their main plan on this one was to: 1. Show supremacy; 2. Get feedback
  • Now they got feedback and are acting based on it
• Duplex takes WaveNet to the next level, adding vocal queues to make the chatbot sound more natural when in a conversation. The result is uncanny, and you can see by the laughs of the crowds at I/O
• Duplex is a reversal of the contact center paradigm
  • Contact center software, chatbots, ML and AI are all designed to get a business better talk with its customers. Usually through context and automation
  • Duplex is all about getting a person to better talk to businesses. First use case is scheduling, but if it succeeds, it won’t be limited to that
  • What’s there to stop Google from reversing it back and putting this at the hands of the small businesses, allowing them to field calls of customers more efficiently?
  • And what happens once you put Duplex in both ends of the call? An AI assistant for a user trying to schedule an appointment with an AI assistant of a business
• When this thing goes to market, Google will have access to many more calls, which will end up improving their own services:
  • An improvement to the accuracy and scenarios Duplex will be relevant for
  • Ability to dynamically modify information based on the content of these calls (it showed an example of how it does that for opening hours on Google Maps during the keynote)
  • Can Google sell back a service to businesses for insights about their contact centers based on people’s requests and the answers they get? Maybe even offer a unique workforce optimization tool that no one else can
• I’d LOVE to see cases where Duplex boches these calls in Google’s field trials. Should be hilarious

You’d like to read what Chad Hart has to write about Duplex as well.

For me, Duplex and Assistant are paving the way to where we are headed with voice assistants, chatbots and AI. Siri, Cortana and Lex seem like laggards here. It will interesting to see how they respond to these advancements.

Current advancements in speech recognition and understanding make it easier than ever to adopt these capabilities into your own products.If you plan on doing anything conversational in nature, look first at the cloud vendors and what they offer. As this topic is wide, no single vendor covers all use cases and capabilities.

While at it, make sure you have access to a data set to be able to train your models when the time comes.

Google Lens (=vision)

Where Google Assistant is all (or mostly) about voice, Google Lens is all about vision.

Google Lens is progressing in its classification capabilities. Google announced the following:

• Lens now recognizes and understands words it “sees”, allowing use cases where you can copy+paste text from a photo – definitely a cool trick
• Lens now handles style matching for clothing, able of bringing suggestions of similar styles
• Lens offers points of interest and real time results by offering on-device ML, coupled with cloud ML

That last one is interesting, and it is where Google has taken the same approach as Amazon did with DeepLens, one that should be rather obvious based on the requirements here:

1. You collect and train datasets in the cloud
2. You run the classification itself on the edge device – or in the cloud

It took it a step further, offering it also programmatically through ML Kit – Google’s answer to Apple’s Core ML and Amazon’s SageMaker.

Here’s a table summarizing the differences between these three offerings:

Google Apple Amazon ML Framework TensorFlow Core ML + converters MXNet & TensorFlow Cloud component Google Firebase none AWS SageMaker Edge component ML Kit Core ML AWS DeepLens Edge device types Android & iOS iOS DeepLens Base use cases

• Image labeling

• Text recognition
• Face detection
• Barcode scanning
• Landmark detection
• Smart reply

Handpicked samples from open source repositories Samples:

• Object detection
• Hot dog not hot dog
• Cat and dog
• Artistic style transfer
• Activity recognition
• Face detection

Proprietary parts Cloud TPUs and productized use cases iOS only AWS ecosystem only Open parts Devices supported Machine learning frameworks Machine learning frameworks

 

Apple Core ML is a machine learning SDK available and optimized for iOS devices by Apple. You feed it with your trained model to it, and it runs on the device.

• It is optimized for iOS and exists nowhere else
• It has converters to all popular machine learning frameworks out there
• It comes with samples from across the internet, pre-converted to Core ML for developers to play with
• It requires the developers to figure out the whole cloud backend on their own

 

AWS DeepLens is the first ML enabled Amazon device. It is built on top of Amazon’s Rekognition and SageMaker cloud offerings.

• It is a specific device that has ML capabilities in it
• It connects to the AWS cloud backend along with its ML capabilities
• It is open to whatever AWS has to offer, but focused on the AWS ecosystem
• It comes with several baked samples for developers to use

 

Google ML Kit is Google’s machine learning solution for mobile devices, and has now launched in beta.

• It runs on both iOS and Android
• It makes use of TensorFlow Lite for the device side and on TensorFlow on the backend
• It is tied into Google Firebase to rely on Google’s cloud for all backend ML requirements
• It comes with real productized use cases and not only samples
• It runs its models both on the device and in the cloud

This started as Google Lens and escalated to an ML Kit explanation.

Need to run ML? You need to think where training the model occurs and where classification takes place. These seem to be split these days between cloud and devices. In many cases, developers are pushing the classification algorithms towards the devices at the edge to gain speed and reduce costs and load on the backend. HWaaS

With everything moving towards the cloud, so does hardware in some sense. While the cloud started from hardware hosting of virtualized Linux machines, we’ve been seeing a migration towards different types of hardware recently:

We’re shifting from general purpose computing done by CPUs towards specialized hardware that fits specific workloads in the form of FPAG.

The FPGA in the illustration above is Google’s TPU. TPU stands for TensorFlow Processing Unit. These are FPGAs that have been designed and optimized to handle the TensorFlow mathematical functions.

TensorFlow is said to be slow on CPUs and GPUs compared to other alternatives, and somehow Google is using it to its advantage:

• It open sourced TensorFlow, making it the most popular machine learning framework out there in a span of less than 3 years
• It is now in its third generation of TPUs on Google Cloud for those who need to train large datasets quickly
• TPUs are out of the reach of Amazon and other cloud providers. It is proprietary hardware designed, hosted and managed by Google, so any performance gains coming from it are left at the hands of Google for its customers to enjoy

Google’s TPUs got their fair share of time at the keynote in the beginning and were stitched throughout the keynote at strategic points:

• Google Lens uses TPUs to offer the real time capabilities that it does
• Waymo makes use of these TPUs to get to autonomous cars

Pichai even spent time boasting large terms like liquid cooling…

It is a miracle that these TPUs aren’t plastered all over the ML Kit landing page.

Going with TensorFlow? You’ll need to decide on the cloud platform you are going to use, especially when it comes to dataset processing and training. Google is working hard to differentiate itself there. Wellbeing

I am assuming you are just as addicted to your smartphone as I am. There are so many jokes, memes, articles and complaints about it that we can no longer ignore it. There are talks about responsibility and its place in large corporations.

Apple and Google are being placed on the spotlight on this one in 2018, and Google took the first step towards a solution. They are doing it in a long term project/theme named “Wellbeing”.

Wellbeing is similar to the AI initiative at Google in my mind. Someone came to the managers and told them one day something like this: “Our products are highly addictive. Apple are getting skewered in the news due to it and we’re next in line. Let’s do something about it to show some leadership and a differentiation versus Apple. Bring me ideas of how we can help our Android users with their addiction. We will take the good ideas and start implementing them”.

Here are a few things that came under Wellbeing, and one that didn’t but should have been:

• Dashboard – Google is adding to Android P an activity dashboard to surface insights to the users on what they do on their smartphones
• YouTube includes a new feature to remind you to take a break when a configured amount of time passes. You can apply the same to other third party apps as well
• Smarter do not disturb feature, coupled with Shush – all in an effort to reduce notifications load and anxiety from the user
• Wind down – switching to grayscale mode when a predetermined time of day arrives

• Pretty Please – Google Assistant can be configured to respond “better” and offer positive reinforcements when asked nicely. This one should help parents make their kids more polite (I know I need it with my kids at home)

In a way, this is the beginning of a long road that I am sure will improve over time. It shows the maturity of mobile platforms.

Not sure how responsibility, accountability and wellbeing like aspects lend themselves to other products. If you are aiming at machine learning, think of the biases in your models – these are getting attention recently as well. Fake News

Under responsibility there’s the whole Fake News of recent years.

While Wellbeing targets mainly Apple, The Google News treatment in the keynote was all about addressing Facebook’s weakness. I am not talking about the recent debacle with Cambridge Analitica – this one and anything else related to user’s data privacy was carefully kept away from the keynote. What was addressed is Fake News, where Google gets way more favorable attention than Facebook (just search Google for “google fake news” and “facebook fake news” and look at the titles of the articles that bubble up – check it also on Bing out of curiosity).

What Google did here is create a new Google New experience. And what is interesting is that it tried to bring something to market that skims nicely between objectivity and personalization – things that don’t often correlate when it comes to opinion and politics. It comes with a new layer of visualization that is more inviting, but most of what it does is rooted in AI (as anything else in this I/O keynote).

Here’s what I took out of it:

• AI is used to decide what are quality sources for certain news topics. They are designed to build trust in the news and to remove the “fake” part out of it
• Personalized news is offered in the “category” level. Google will surface topics that interest you
• Next to personalized news, there’s local news as well as trending news, which gets surfaced, probably without personalization though the choice of topics is most probably machine learning driven
• Introduced Newscast – a presentation layer of a topic, enabling readers to get the gist of a topic and later drill down if they wish in what Google calls Full Coverage – an unfiltered view of an event – in an unpersonalized way

One more thing Google did? Emphasized that they are working with publishers on subscriptions, being publisher-friendly, where Facebook is… er… not. Will this hold water and help publishers enough? Time will tell.

AI and Machine Learning lends themselves well to this approach. It ends up being a mixture of personalization, trending and other capabilities that are surfaced when it comes to news. Can you see similar approaches suitable for your product offering? Smart Displays

Smart displays are a rather new category. Besides Android as an operating system for smartphones and the Waymo AI piece, there was no other device featured in the keynote.

Google Home wasn’t mentioned, but Smart Displays actually got their fair share of minutes in the keynote. The only reason I see for it is that it is coupled nicely with the Google Assistant.

The two features mentioned that are relevant?

• It can now show visuals that relate to what goes on in the voice channel
  • This is similar in a way to what MindMeld tried doing years back, before its Cisco acquisition
  • The main difference is that this involves a person and a chatbot. Adding a visual element makes a lot of sense and can be used to enhance the experience
• It offers rich and interactive responses, which goes hand in hand with the visuals part of it

I am unsure why Google gave smart displays the prominence it did at Google I/O. I really have no good explanation for it, besides being a new device category where Apple isn’t operating at all yet – and where Amazon Alexa poses a threat to Google Home.

Android P

10 years in, and Android P was introduced.

There were two types of changes mentioned here: smarts and polish.

Smarts was all about AI (but you knew that already). It included:

• • Adaptive Battery
  • Adaptive Brightness
  • ML Kit (see the Lens section above)

Polish included:

• App Actions and Slices, bot offering faster and better opportunities for apps to interact with users outside of the app itself
• UI/UX changes all around that are just part of the gradual evolution of Android

There was really not much to say about Android P. At least not after counting all the AI work that Google has been doing everywhere anyway.

App Actions and Slices are important if you develop Android Apps. ML Kit is where the true value is and it works on both Android and iOS – explore it first. Google Maps

Google Maps was given the stage at the keynote. It is an important application and getting more so as time goes by.

Google Maps is probably our 4th search destination:

1. Google Search
2. Google Assistant
3. YouTube
4. Google Maps

This is where people look for information these days.

In Search Google has been second to none for years. It wasn’t even part of the keynote.

Google Assistant was front and center in this keynote, most probably superior to its competitors (Siri, Cortana and Lex).

YouTube is THE destination for videos, with Facebook there, but with other worries at this point in time. It is also safe to say that younger generations and more visual audiences search YouTube more often than they do anything else.

Maps is where people search to get from one place to another, and probably searching even more these days – more abstract searches.

In a recent trip to the US, I made quite a few searches that were open ended on Google Maps and was quite impressed with the results. Google is taking this a step further, adding four important pillars to it:

1. Smarts
2. Personalization
3. Collaboration
4. Augmented Reality

Smarts comes from its ML work. Things like estimating arrival times, more commune alternatives (they’ve added motorcycle routes and estimates for example), etc.

Personalization was added by the introduction of a recommendation engine to Maps. Mostly around restaurants and points of interest. Google Maps can now actively recommend places you are more likely to like based on your past preferences.

On the collaboration front, Google is taking its first steps by adding the ability to share locations with friends so you can reach out a decision on a place to go to together.

AR was about improving walking directions and “fixing” the small gripes with maps around orienting yourself with that blue arrow shown on the map when you start navigating.

Where are we headed?

That’s the big question I guess.

More machine learning and AI. Expect Google I/O 2019 to be on the same theme.

If you don’t have it in your roadmap, time to see how to fit it in.

The post Google I/O 2018 and the Future of Computing appeared first on BlogGeek.me.