Top news stories for Episode 15 (October 19, 2017):
1) Voicebot.ai: Smart Speaker Sales Grow 300% over 2017, to 24M Globally
2) Voicebot.ai: The Amazon Echo "Connected Home Experience" now available in 10 Kohl's stores; rolling out to 72 more
3) Voicebot.ai: Samsung's Bixby claims over 10M users worldwide
4) Voicebot.ai: Alibaba Tmall Genie To Go Into 100,000 Marriott Hotel Rooms (Across China)
5) Voicebot.ai: Alexa can now differentiate between user voices, opening the door to personalized content
6) Voicebot.ai: Garmin introduces Garmin Speak, a voice-first device designed for in-car use
7) 9to5Mac: Apple explains "Hey, Siri" in latest post to its Machine Learning Journal
This Week In Voice available via:
Voicebot.AI's article summarizing this episode of This Week In Voice is available here.
Panel for Episode 15 (October 19, 2017):
Bret Kinsella is publisher and editor of Voicebot.ai, home to original opinion, commentary, and analysis on all things voice web. Check out the page for the latest news and to subscribe to the weekly newsletter.

Transcript
Bradley Metrock: [00:00:10] Hi and welcome back to This Week In Voice, Episode 15, for October 19, 2017. My name is Bradley Metrock. I'm CEO of a company called Score Publishing based here in Nashville, Tennessee. Our sponsor for This Week In Voice is VoiceXP, blazing the trail in voice technology. VoiceXP is taking the lead in developing Alexa skills for some of the best brands in the world. With VoiceXP, all you have to do is say it, to revolutionize your marketing strategy. As I say every episode, if you have not checked them out, you are doing yourself a disservice. Head on over to www.voicexp.com. Pause the podcast, go on over there, you'll be glad you did. I am very pleased to have Bret Kinsella as our guest today. I am thrilled about this. Bret, say hello.
Bret Kinsella: [00:01:08] Hello everybody out there in #VoiceFirst land.
Bradley Metrock: [00:01:12] Bret, thank you very much for making the time. Bret is the editor and publisher of Voicebot.ai, which is a fantastic news and commentary site for all things voice web. Do I have that right?
Bret Kinsella: [00:01:27] Yeah I think that's right. We say, "Giving voice to a revolution."
Bradley Metrock: [00:01:32] There you go. Perfect. Give us a little bit of background, Bret, on how you decided to create the site and how you manage the day-to-day curation and creation of the articles that are on there.
Bret Kinsella: [00:01:47] That's great. Well, probably the best way to think about it is a passion project. About a year and a half ago, I was asked by a client...I have a marketing agency that I've run for a number of years. We work with early stage venture-backed startups. One of my clients asked me to do some research into this space and I was frustrated when I started doing the research because I found that the information was very thin. There were a few things about devices, and their reporting was fine as far as it went but it really didn't explain the industry very much. And ultimately I did my research, I found the information I needed. And at the end of that I said, "Oh geez, I've seen this before." Because I was working in the Internet and the rise of the web in the mid 1990s and I saw so many hallmarks... there were complete parallels between what was going on here and what was going on then. I said, "Well, this is really important."
Bret Kinsella: About the same time I was asked to write an article for Huffington Post and Advertising Week and I put it out there. And I really got a tremendous amount of response. People asked me where I got the information and what to do next. So because I'm in marketing, I decided I would just put up a website to share all of the links and sources that I had. And I thought maybe I would blog once or twice a week on it. And things kind of blew up from there because that was September of last year, September 15th. The first week I put it up, I want to say, the Echo Dot was released. Two weeks later the announcement around Google Home. Maybe the week after the Viv acquisition by Samsung, which we're going to talk about today. So because of all those things happening it turned from something that I thought I would just put a little bit information up once or twice a week into publishing 60 times a month and all these other things. And so it's really been tremendous. I've had a great time meeting a lot of entrepreneurs in the industry. A lot of really, really smart people working in both the voice and AI space and I do cover both, not just voice. It's got a heavy slant towards voice, but I do a lot of AI as well.
Bret Kinsella: But just meeting so many tremendous people and seeing the excitement and enthusiasm. That really fuels it. And it's not just me. So Ava Mutchler and I have worked together for a number of years. She is the associate editor on the site. A lot of the articles people have read have been written by Ava. She's probably done at least 50 percent of the articles on the site. We expanded into a podcast as well because a lot of the people I was talking to... I just couldn't do justice to their stories in twelve hundred words, right? And I have to say I'm thrilled about what you've been doing because I think This Week In Voice is great. I'm an avid listener and I like what you're doing with VoiceFirst FM. Really, it's been a great resource to the industry. So I really like the way this industry is growing up and I've really been excited about what's happening.
Bradley Metrock: [00:04:56] I appreciate that. Thank you for sharing that and I appreciate that. I view Voicebot.ai as essential reading. It's so unique that it really doesn't matter if you don't go to the site because you're just going to see people linking to articles from it and you'll find your way around anyway. As I was saying before the show it's a very healthy side of the ecosystem. Something like what you're doing exists but the quality that you are putting into it shows, and so there's a lot of people who appreciate what you're doing and from the standpoint of VoiceFirst.FM I think it's two complementary resources. You guys are outsourcing so much information and you're presenting such quality news and insight and analysis. VoiceFirst.FM... really is just a mirror held up to reflect back at the sector. We allow people to tell their stories. So I think it's great. I think VoiceFirst.FM and Voicebot.ai are very complementary and look forward to continuing to work with the brand. I'm happy with what you're doing and appreciate you setting the time aside today.
Bret Kinsella: [00:06:22] Agreed and I'm thrilled to be here.
Bradley Metrock: [00:06:25] So with that let's get to the news. Our first story this week - and by the way as you look at the news stories of the week, six of the seven are Voicebot.ai stories. I really thought that would be appropriate with Bret being our guest, which we very much appreciate. And as a side note, we have a Voicebot.ai story of the week, each week, on This Week In Voice that highlights a particular story of interest that Bret and his team have covered. The first story this week is “Smart Speaker Sales Grow 300% over 2017, to 24 Million Globally.” And Bret, my question for you is: when you look at the growth of smart speakers...it was just earlier this week or late last week where I saw somebody was presenting... some venture capitalist...my days have been blurring together lately. I don't remember the person's name nor do I remember where he was speaking, but a venture capitalist or somebody savvy in the field was making the statement that he thought smart speakers were going to grow very rapidly over the next two years and then basically completely die off as they are equally rapidly integrated into existing appliances and other existing context. As you look at the story of smart speaker sales growing, what is your takeaway? Do you think something like that's going to happen, or do you think we're just going to see more and more growth longer term beyond the next couple of years?
Bret Kinsella: [00:08:04] Well, the only thing I know about forecasts first and foremost is that they're always wrong. Sometimes they're low, sometimes they're high, but they're never quite right. And the thing that I have found ... the empirical evidence would suggest that in this space, all forecasts are inadequate and that they under-report what's going to happen. So I would say this is true in this article. I think the strategy analytics guys do a wonderful job. Several others in the industry who do ... and I like the methodology that they've done. I think they are good on the smart speaker side but they are light in thinking about all the other access points, because that wasn't the purpose of their study. And so I'm somewhat sympathetic to the reference you made around this idea that they're going to trail off and that they're going to be embedded into other devices.
Bret Kinsella: I think for sure the second part of that statement is correct. We're already seeing it. I mean, you're seeing it in cars. You're seeing it in appliances. But trail off and die... I suspect is not quite right because there's a long way to run. This is not two years. You know, we're looking at something with like probably 15 percent household penetration in the U.S. over the next year. I think that investors might have only been looking at the U.S., but even there you've got another 5x growth likely over the next three years beyond this year just to get into households, and that's before you have more ... that's before businesses start putting them in every office. I think that's not quite right. But I do agree with the fact that we're going to find microphones just about everywhere very soon. That's going to take a little longer than a lot of people think. But it's going to be there just because the utility is so high.
Bret Kinsella: Now the question is then, what are the use cases? And if you look at most of the other data, the use cases are today predominantly transactional utilitarian solutions like conversions of metrics, math, weather, those types of things, access your calendar maybe, and then music and audio content. And when we get to that... audio content is far and away the longest in terms of time spent with the devices and also usually the highest in frequency or in the top 3 in frequency of use in any studies I've seen. So what does that mean? Are smart speakers going to go away? Are we just going to be listening to the radio through our appliance? Not unless the appliance delivers a better speaker. I think there's still a lot of utility for these devices. They're really inexpensive now as we all know. We can put them in lots of rooms. So two years, no chance. I think we'll see ... it will be asymptotic like anything else as you get saturation and the market's growth rate will decelerate. But absolutely we're looking at another five to seven years of really strong growth.
Bradley Metrock: [00:11:19] That's great. It's like turning on the firehose of information like I thought it might be. That's excellent. Let me ask you just a sort of a related side note. It comes up from time to time on this show and others about the Homepod at Apple. Just generally... smart speaker sales grew 300 percent over 2017. Obviously and as your article indicates, the lion's share of that is Amazon, but there's other opportunities, as the market figures itself out and the use cases become more and more known. In your opinion ... so Apple's got 500,000 Homepods to sell. That was pretty widely reported, as well as on your site too. Tell me what you think about Apple's ability to compete in the smart speaker space with the Homepod and beyond in the context that the whole ocean is rising.
Bret Kinsella: [00:12:22] Yeah, it's a great question and I've reported on that 500,000 number and others to sort of give your listeners an insight into something ... a conversation I had with another analyst in the industry last week and they're expecting will be more like a million because they think that some of the supply chain issues have already started to get sorted out and that they can push a million. And so what does that mean for Homepod? They're going to sell in three countries: U.S., Canada, and Australia. That's not a lot of units for three countries like that, particularly with the US being 300 million people and you think about the number of iPhone users in the country. They're going to have no problem selling 500,000 or a million. I just don't see that that's going to be an issue even at the price point because the numbers are not a lot. Now, could they sell 5 million? I'd be skeptical at that price point, just because there are so many alternatives even with high-end audio that are available out there. However, as anybody who knows Apple products knows, when you have Apple products and you add a new Apple product to that ecosystem, it's a good experience. The cross-product integration is excellent.
Bret Kinsella: And I think the thing that gets the most play is Airpods. I've written extensively about Airpods and I've used them for a while. They're really wonderful. The issue for Apple is not the smart speaker, it's not playing music; the issue for Apple is the very limited capabilities of Siri and that comes from a couple of sources. But Siri, we all know, can do some of the utilitarian transactional things that I talked about earlier: it can tell you the time, it can do conversions, those types of things. It could even get you music if you're okay listening to Apple Music. The thing it can't do ... it really doesn't have the broad domain capabilities that we've come to expect now with Alexa and Google Assistant, and there are two constraints there. One is that there's actually only nine domains that Siri allows people to develop against. So there's a lot of things that you might want to do that there's really no way for a developer to integrate with through the SDK. The second thing is: in order to develop for that platform today you actually have to build an iOS app first and then add the voice on top of that. And building an iOS app is more complicated by a significant degree than building a Google Action or Alexa Skill.
Bret Kinsella: And so that's a significant barrier for a lot of people who've already built popular, useful types of things in the voice space; for them to think about importing this to Apple is a big lift. And we've already seen the friction between the Amazon ecosystem and the Google ecosystem, where Google's more complicated. There are some tools like api.ai, or it's now called Dialogue Flow, that make that a little bit easier. But still it's more complex than building Alexa Skills. So we've got all these things that are going on with Apple that are really longer term issues for them. In the near term I don't think it is. People are going to buy it, they're going to listen to music, they're going to have some basic Siri capabilities, and they'll be happy. And it's not too late. But the limitations of Siri are really a problem and I don't see Apple addressing them until WWDC in June of next year, which means that they're not really going to be available for consumers until about this time next year. So we've got another year to wait, and by that time I'm guessing we'll be looking at 30-35 percent of households in the U.S. that will have it. They're going to be struggling a little bit and playing catch up in the U.S., but if they don't have a low price option that's going to be a big problem.
Bret Kinsella: However, all those things being said, the world is very large, and Apple has a presence all around the world with the iPhone and with all of their products. And most of the world is going to be very early in the adoption cycle at that point. If Apple really has an aggressive 12 months between now and next fall, they're going to be a key global player and they could turn out to be one of the top two global players in the end in this space. And they've got a lot of assets, not just the iPhone distribution and the broad product portfolio, but they also have language models because Siri recognizes lots of languages people don't know - I think it's over 30 now. And at the very least they've got the ASR aspect of it, the automated speech recognition, so they can understand and they can do that. The speech-to-text, they can do the speech-to-text. What they don't really have is the NLU, and they don't have the NLU necessarily tuned to all those localities or those languages. So that would be a place that I would see them investing heavily. That's where Google is sort of in a race to get there before Apple, because there's a lot of players out there.
Bret Kinsella: But we should give a shout-out to Adam Marchick at VoiceLabs, who I know you've interviewed. Because in January he came out with his estimate of 24.5 million units to be sold this year. I think when you look at something like what Strategy Analytics came out with, it nails that number. Of course not just smart speakers, he said. #VoiceFirst interfaces, you know. We might be cutting hairs here on what people's models were. But the fact is a lot of people said, "Oh, no, that seems kind of crazy." It was widely reported at the time because people liked it. They thought it was a sensational headline but I think there was a lot of skepticism, and frankly at the end of this year I think we're going to find that Adam was a little too conservative. But he was the first one out there to say, "Hey everybody out here who's thinking that this might only be a 10 or 12 million unit year ... you're way off."
Bradley Metrock: [00:18:31] Yeah, and I saw that mentioned in your article. That is great to give him a shout-out and to be that close in January to what people are projecting in October for the end of the year. You did something right. Which is great.
Bradley Metrock: The second story is about Amazon rolling into Kohl's: “The Amazon Echo ‘Connected Home Experience’ Now Available in 10 Kohl's Stores,” and the Voicebot.ai story on this lists the locations of the stores, which I thought was really interesting and a nice addition. And then it's rolling out to 72 more Kohl's over the near future. So Bret, My wife loves Kohl's. I mean she loves Kohl's. She also loves Amazon and it was some point on the show. Someone made the connection between ... I think it was Brian Roemmele who made the connection between Amazon and Kohl's and now there's a pretty healthy intersection on the Venn diagram of their customers. So this makes perfect sense. Do you think that this is leading to an inevitable acquisition of Kohl's by Amazon? And what else should we take away from this story?
Bret Kinsella: [00:19:56] Well the answer is no, I don't think it's an inevitable acquisition, but I think that's clearly on the table. And I'll say just about the Venn diagram, I'd say that Amazon's user base is so broad now that their user base overlaps with every other retailer’s user base.
Bradley Metrock: [00:20:13] True. Good point.
Bret Kinsella: [00:20:15] So I do think it's kind of interesting here. Amazon's apparel business has been growing very fast. And that's what Kohl's primarily is known for, the apparel. Ava Mutchler, who I work with, has written about this a few times on Voicebot.ai and she used to work in the fashion industry, so she has particular insight there. But that's not why Amazon is there. It would make sense at some point because apparel is such an "experience buy" ... people like to touch it, they like to know the fabrics, they like to try it on, that eventually Amazon is going to run up against a wall there in terms of how much they can sell online and in that type of experience. But I think the number one thing here is not that yet. That's like the option value down the road. So first of all I think it's a great opportunity for both companies and I pointed that out a lot of the analysts say, "Geez, Kohl's, you're crazy." But I don't think so.
Bret Kinsella: First of all, Amazon doesn't have a place you can go to visit it today. So now you can go visit Amazon at Kohl's. I think that's a tremendous foot traffic opportunity. And I've only been in a Kohl's store a couple of times and I've only done a little work myself in retail. But I can imagine the promotional displays, the discounts that are going to be circling the connected home cutout within Kohl's. And this isn't the first time Kohl's has done this, by the way. Think about Macy's department stores. You've got all these different brands who have their sections at the store. And I think it was last year that Kohl's started with Under Armour and there's one other brand maybe that they've done this with and they've had some good success with that. They basically give that brand a part of their store. And in this case, home electronics buyers wouldn't naturally walk into Kohl's … but we all wear clothes, so we go in and we want to look at the Echo Show because we've never seen it before and then they say, "Wow, there's new sweaters that I could buy for Christmas!" Right? And what I think is brilliant about this, too, is there are two reasons for people to go to these stores now for Amazon.
Bret Kinsella: One is that today you can't really look and feel an Echo anywhere in retail if you want to look at it and test drive it before you buy it. Now there's a place you can go. 72 stores is not a lot. It's not a lot of coverage in the U.S. You talk about Walmart with over 4,000 stores. It's much different. What you have to keep in mind that Google Home is in all of those Walmarts. Google Home is in all the Targets. And so that's an advantage that Amazon's primary current competitor in that space has. And so this at least gives them this opportunity to have some parity in that space. You talk about this whole idea of "showrooming" that people go to Best Buy and they buy an Amazon Echo, those types of things. Well, now they can actually go and test drive some of these products that don't exist in many other retail locations. There are a few places you can get it but it tends to be fairly limited. I think that's a tremendous opportunity.
Bret Kinsella: The second way people are coming is if you buy something and you want to return it, you can just print out the label and go to the Post Office and send it back to Amazon, or you can just go to Kohl's give them your stuff. They'll package it up and return it for you. And while you're there maybe you buy an Amazon product in that showroom, or maybe you'll look across the aisle and you'll see a new sweater that you'll want to buy for the holidays for one of your relatives. So I think this is a really tremendous thing. And I'll just close on this one point. Amazon is really big in online commerce, but online commerce is about one-twelfth of total shopping in the U.S. So in order for them to continue to grow and to grow at a rate faster than online commerce, which has been decelerating a little bit in terms of its percent of total shopping … they need to find a bigger market. And one of the ways to find a bigger market is … all of a sudden they're available in physical retail. So what they've done with Kohl's is, it starts to introduce them to that larger retail market, that physical shopping experience, which is 12 times larger than the online shopping market that they operate in today.
Bradley Metrock: [00:24:28] Yeah, that's great information. I owned a retail business from 2008 to 2013 when I sold it. And it's fascinating to me to see Amazon ... most people don't realize, although I'm sure that you do, that it's just a totally different dynamic in a physical store. There's impulse buying and specifically impulse buying is the thing on my mind. Impulse buying that takes place within the walls of a physical location that you can never replicate. At least with the current online tools. The Amazon shopping experience, as good as it is and how it is facilitates discoverability and everything ... there is some magic that takes place in a physical location. And Amazon clearly understands that. And like I said, my wife loves Kohl's. She talks about Kohl's Cash and I would normally never know what Kohl's Cash is. But thanks to her, I do, and she's extremely loyal to that company. For what reasons, I don't fully understand, but it's good pricing, it's the products she's looking for, it's the loyalty programs and all that.
Bret Kinsella: [00:25:52] Yeah, I'll say one other thing, too, that I think Amazon has the potential to tap into - online shopping is a personal and individual experience. Physical retail shopping is often a social experience.
Bradley Metrock: [00:26:06] Sure.
Bret Kinsella: [00:26:07] And so you know sometimes it's family members, sometimes it's friends, relatives, whatever … but people go together to shop very often. And that in itself expands a whole new world of opportunities for Amazon to introduce people to a new type of shopping experience.
Bradley Metrock: [00:26:26] I did not have any idea. Maybe you missed in the article but I missed it. That's entirely possible. But the idea that you can take returns to Kohl's and they'll box them up and handle that ... that's what you said, right?
Bret Kinsella: [00:26:39] Yeah, that's true. In fact, most of the reporting on this focused on that angle, and I think it's an important angle. I just ... the Voicebot.ai audience is more interested that you can go buy the devices there. Yeah, I did mention that and I think that's one of the key things, and now there's two reasons to go there to see Amazon s. And that's just a foot traffic magnet potentially for Kohl's. I'm fully expecting Kohl's to have a good Q4 and that's in part driven by this.
Bradley Metrock: [00:27:08] If I could walk into Kohl's and do an Amazon return with somebody facilitating that … I may walk into Kohl's. So that’s a huge aspect of that. I appreciate you mentioning that. Amazon, they're doing good work and it's cool that they are identifying partners ... for as global and as much of a juggernaut as they are. They're finding partners that can add immense value to what they're doing. So that's great.
Bradley Metrock: Moving on to story number three: “Samsung's Bixby Claims Over 10 Million Users Worldwide,” in an executive presentation that was made recently, which is a lot of users. It's not as many as perhaps people using Siri or perhaps using Alexa, but it's still quite a bit. Bret, I want to ask you... it's sort of a theme of several of these that's been mentioned on this show, This Week In Voice, a bunch and sort of a thought that's out there that there's still a lot of development. This market is still new, as you mentioned earlier, and it's still immature and there's a lot of growing to do. And as that happens the leaders can change in a hurry. So my question for you is ... Bixby's got 10 million users today worldwide ... how do you see Samsung fitting in globally over the short term? Where do you think their low-hanging fruit is in the global marketplace? And do you see them being able to compete effectively in the United States?
Bret Kinsella: [00:28:54] I think Samsung and Bixby are an interesting story. It's hard to say right now what their prospects are going to be. But let's break it down a little bit. First of all, the 10 million you said, that's enough of a user base for them to have a foothold to grow from. And that's for a couple of reasons: one, we're forming habits and consumers are starting to have experiences with voice assistants, and so familiarity will be there for at least a certain segment of the population. The second thing is the training data that they'll get from that which will ... they'll have enough users on it that they'll start getting data, so that Bixby AI will continue to improve. And then ultimately ... the biggest asset that the company has is that it's always the number one or two global smartphone manufacturer, so they can they can put Bixby in front of a lot of people, and then it's just a matter of making sure that the utility is better than the Google Assistant, which is also on those same devices. And here's where it might come down to technology in those instances.
Bret Kinsella: Bixby 2.0 is the first that we will see of Viv, which was the acquisition last September which we mentioned earlier in the show and that was founded by the people who were actually among the original team that founded Siri. And it was sold to Apple, and then a few years ago they left and started their voice startup and created Viv, which is like an Alexa. In fact the parallel that is probably better is Soundhound's Hound for a couple of reasons. But here we have some people who’ve been in voice for a long time. Siri came out … it was announced in 2010 and it was first unveiled in 2011. They know what they're doing. One of the things that they talk about is compound requests and multi-service requests. They don't use those terms but there would be a way that most people in this industry would think about it; a compound request is to say something like, "Open this. Tell me this and do that." Right? So there's three things that you're asking for. And two after you engage the voice assistant.
Bret Kinsella: If you look at what we do with Alexa and Google assistant, they're not compound requests. They tend to be single shot requests. So this idea that you can ask for more than one thing at a time and sort of simplify the interaction is an important step forward, and one that Alexa and Google Assistant need to get better at. The only thing is, this concept of multi-service requests … which is that your skill or voice app doesn't need to be able to do everything ... that someone could invoke your voice app, called it like a service, and then they could access other services that have information or capabilities beyond yours. So for example, you could open it by asking to look up your cell phone bill and also pay it using Venmo or Square or something like that. So what you're doing is, you're doing a lookup and you're doing a pay on one command. Those are the types of things where we're going to see over time ... You'll probably see it first in travel and some of these other areas where you've got this deep domain expertise and no one voice app is going to be able to answer all the questions. But if they have a capability, they can reach out to other services that have already captured that domain experience, so that they could have these interlocking sets of voice apps … unlike the iOS and Android world, that's only been available from app-to-app connectivity, for just a couple of years now. And it opened things up quite a bit but it's still highly restrictive. This seems like a much more fluid way for us as consumers to have essentially a boundary-less voice app ecosystem where we go to the places where we want things to be executed, but they don't have to do everything themselves, they can use other services. And that's really important from a technology standpoint.
Bradley Metrock: [00:33:05] I think everybody hopes that that is the way the market continues to move and that is why that Alexa and Cortana partnership was so significant … because if Amazon and Microsoft can find a way to play nicely together, then there's no reason why any other two players can't. You know what I mean?
Bret Kinsella: [00:33:27] Yeah, I think that makes a lot of sense and it's even within the voice assistant platform allowing the third party skills and voice apps to be able to collaborate. That's the model that Soundhound's Hound is using today, which I think is a really interesting way to think about it. Once you're in that ecosystem and you might be using Hound as your point of access but you can go to one application that's using Hound, then that application can then access information capabilities from other applications in that area to perform more complex or valuable services. Ultimately … what we would have called that the in technology world is a service oriented architecture. You think about Web services. No matter what your point of entry is, you can get other things outside of that point of entry whenever you need it, and then the utility goes up. When you talk about a true assistant … if you had a butler, the butler wouldn't be able to do just one thing. He's not just going to pour you tea. He could also go to the store, right? Or he could call a car, or he could get something to come by to fix the electricity, or something like that. I think that's where we need to head with this. And right now the existing platforms that are popular don't allow that type of interaction. And in this sense … we haven't seen that yet. We have to see what the architecture is. But it will be a tremendous step forward if they do. And if that's the case, and they can get developers to build to it, which is a whole other story, then we could see a really robust set of capabilities on that platform that are not replicated on the other platforms.
Bradley Metrock: [00:35:10] Yeah. I'm in complete agreement. Just imagine if you went on the Safari browser and you typed Amazon.com and it returned and it said, "No thanks. Go use another browser or another computer if you want to access that website." It would be absurd, and I think we're going to get to that point very quickly, if we're not already there with voice. And the other thing that I think is worth mentioning here too … it's not to be understated, the fact that you've got entirely new generations of humans who are growing up with these computers, with Alexa, with Siri, and becoming accustomed to doing stuff with voice and with young people, they're not all that interested in big corporate strategic “strategery.” You know, they just want stuff to work. So having an open architecture and flexible passageways in and out of different voice assistants and ecosystems, I think is going to be fundamental very soon. So that's great.
Bradley Metrock: Moving on to story number four, “Alibaba Tmall Genie to go into 100,000 Marriott Hotel Rooms (Across China).” I don't know that much about Alibaba, but Brian has said a bunch, and different guests on the show have known differing amounts about them. Obviously, they're a huge company. When I look at this, it just screams to me … there's no place that voice is not going to go. Every business is going to have to have a voice application. (And if you need one, call VoiceXP, shameless plug.) But not only is every business going to need a voice skill or application, if not many ... but then there's a whole hardware strategy that every single business that exists in the organization has to think through of what hardware should they be providing to facilitate voice and voice interactions. And so my question for you, Bret, is: what should we be taking away from this story? Perhaps you agree with what I just said, or even beyond that? Yeah, I'll leave it there. What should we be taking away from this story?
Bret Kinsella: [00:37:51] Fair enough. I think your initial response is a good one. Your initial inclination there, it shows that we're going to find microphones all around us in the environments we inhabit because there's utility there and people are already seeing that. Really, the angle I take on it, though ... I'm particularly interested in the voice assistant smart speaker battle in China. Essentially, just to break it down really quickly: you've got Alibaba, and the best analogy is that they're the Amazon of China. They dominate online commerce in the country. They've got other things. They've got an AWS type of service as well. They're in taxis, they're like an Uber competitor. Interestingly enough, their two key competitors are Baidu, which is like the Google of China, who has DuerOS. They've been a little slower to roll out devices, but they do have a voice assistant called Duer. And then you've got Tencent which is more like a media company, closer to a combination of Facebook and Netflix. And they have a big online commerce group subsidiary called JD.com and their smart speaker is called LingLong DingDong. For those of you haven't heard about it, it's kind of a catchy phrase and it has a specific meaning in Mandarin which I don't recall, but I bring that up... Then there's Huawei as well, which is the number four handset maker globally. They're also playing there.
Bret Kinsella: So I look at what we're seeing with those companies is very similar to what we're seeing in the U.S., where you have Google up against Amazon up against Microsoft up against Apple. You've got some big global players, and they've got a lot of cash. But the important thing right now is that what you see in them. It's kind of like China and the rest of the world. Everything's bigger in China, they're so focused on that domestic market … a hundred thousand is a lot of hotel rooms, but there's a lot of people in China. There's a lot of travel there. So that's just scratching the surface there. And all of those players that are offering Mandarin solutions will eventually offer solutions probably for Cantonese and some other variants. The question is: are they going to get out of the Chinese language? And I'm not sure that they will. They're really so focused on just battling each other to set up that competitive space in China. And I think this gives Alibaba an edge. They already have an edge, because they’ve got the retail distribution channel. But I think this is just another way where they're going to introduce people. They're going to get maybe their first experience with a smart speaker; might be in a hotel room, and then they might buy it and form some habits early on. So I think this could be big for them, but we’re talking about it being early days here in the United States. It's way earlier there.
Bradley Metrock: [00:41:19] It was Mandy Chen, who was on this program, who had mentioned or sort of connected the dots in a way I had not thought about, in saying that the rise of voice in China specifically, that was what she was talking about, directly related in a large degree to the complexity of the written language.
Bret Kinsella: [00:41:47] Yeah, that's absolutely right.
Bradley Metrock: [00:41:49] Yeah. And that is very interesting. And of course that just makes common sense that that would be true. But it opens the door to taking a look at all the different individual geographies across the world, and assessing how much or how little they might be interested at least at first into getting into voice. But I just thought that was very interesting in how that's pushing that market forward. But I appreciate your analysis on that and we will keep an eye on how that market continues to evolve.
Bradley Metrock: The next two stories, I'm actually going to combine. Story number five this week is “Alexa Can Now Differentiate between User Voices,” which is something that Google Home has been able to do for some time. And then story number six is that “Garmin Has Introduced Garmin Speak,” which is essentially a piece of hardware that is Alexa-enabled that goes in the car. And rather than talk about those two things individually, I want to sort of open the door to a discussion. You know as well as I do, Bret, there's been tons and tons and tons and tons of news as far as Amazon has been concerned this year. I wanted to simply ask you this question. In your opinion, what is the best thing that Amazon has done this year? And what is the lowest hanging fruit, the thing that you wish that they would do immediately?
Bret Kinsella: [00:43:30] The best thing they've done this year is Echo Show. I don't think there's any question about that. Introducing a true multi-modal visual voice experience opened people's eyes to what's going on in the world. Plus, I've written about this, where essentially they introduced a rotating billboard into my kitchen, which has all sorts of tremendous applications as you can imagine.
Bradley Metrock: [00:43:54] Sure.
Bret Kinsella: [00:43:55] The thing that I think that I would most like to see them do ... and I'm really taking probably more of a supply side view this as opposed to a demand side because I do work in this space and I work with some product companies. But it's kind of astonishing to me that they do not allow companies with Skills to sell their products directly from the Skill, they actually have to go out of the Skill into Amazon. It's still in Alexa but they sort of leave that bubble of the Skill so I invest this time to get someone to try my skill. They're interested in my product. And then in order for me to convert that transaction, I have to ask them to leave. Essentially how it is, I have to say, "Alexa, order this thing." I think that's a big problem because voice commerce is such a big deal, and I think this idea of user voices is key to that, that’s one of the key things around it. I think there are some other things … convenience and parity with Google … but this idea that I should be able to order by voice because it's me, I think is important. And I think we should just be able to go to these destinations which are specific to a brand and be able to order from directly inside that.
Bret Kinsella: And I expect that to come soon, and I'm kind of surprised it's not there now, but I think that's a really essential thing for the healthy development of that ecosystem. And in part of that's because we talk in this industry a lot about monetization. I know you've talked about it a lot in this show and your other shows related to voice. That's a really quick way for some people to say, "Here's my ROI. I can build something and it can be high quality and I can sell from within it." It's also, even if you don't have your own products, it's also a way for other people to start to monetize their Skills where they can sell other people's products or solutions that are complementary to what they are from within that Skill. And I suspect that that's maybe the most fundamental, mechanical change that needs to be made in this market. And Google's already there. They've got to preview of this for developers who can start using this. Amazon needs to get there.
Bradley Metrock: [00:46:07] Yeah, I am completely in agreement with you on the Echo Show. I've mentioned that before on this show.
Bret Kinsella: [00:46:15] I know you're a fan.
Bradley Metrock: [00:46:16] Yeah. How can you not be? I've mentioned on one episode of the show, and of course I don't remember which one it is, and I might have mentioned on the VoiceFirst Roundtable as well at some point … what Amazon is doing right now is legendary business performance. It's what's going to be written about by perhaps you or somebody else or me or somebody, some observer. For years to come there will be books written about it. It's historic. The event they had recently where they announced seven or eight different products … one of which was the Echo Buttons, which are phenomenal. I really liked that they're doing that. They're so in command of this market and leading the marketplace that it's like an incredible musician, it's like Jimi Hendrix or Prince. Like they know the guitar so well that you're about to see them do some stuff that other people could only dream of. You know I mean? It's operatic.
Bret Kinsella: [00:47:22] I think part of it is that they worked for two ... well, they worked for more than two years. But you think about the two year head start, it's probably really like a four year head start, on most of their competitors, and a lot of sorting through a lot of the messy details already happened. So now they're just sort of in this execution and they know that there are certain things they want to try because even if they fail, it doesn't matter because it's all part of this grand strategy. Whereas I think what we're seeing with Google and maybe to a lesser extent Microsoft and I guess we will be seeing with Apple … it's hard to see anything because they're sort of opaque. We're seeing the messiness of them working through these processes that Amazon was doing before when no one was paying attention.
Bradley Metrock: [00:48:10] Sure. So there's two types... I mean most people are very unaware of all the different things that Amazon is doing, because how could you possibly be? I mean there are things, for as much attention as you and I pay to the marketplace, that catch us by surprise or that we don't know. But for the layperson it's just like they only encounter things as they use them. The breakneck speed that Amazon has had in the marketplace that's too much for them to keep up with … it's fun to watch what Amazon is doing. And Amazon, yes, their employees definitely seem to have license to fail as they pursue different things that they think are good ideas. But it's also very, very helpful that Amazon is such a trusted brand. I'm talking about in the U.S., obviously.
Bradley Metrock: But Amazon has gotten away with things that other companies would not get away with because they are who they are and they're so good at customer service and their pricing, their dynamic pricing models on the retail side favor the customer. It's very interesting the interplay between the parts of the business. So the other thing, to put a button on this, is that yeah I think the monetization needs to come, as I've talked about before. And so just to flesh this out, I want to get you on the record. Are you in favor of Amazon having an App Store-style monetization scheme for the likes of Skill marketplace, or not?
Bret Kinsella: [00:49:48] I'm kind of indifferent in terms of whether it's exactly the App Store model, but I'm certainly not opposed to that. I think being able to sell - if you think of what the App Store model is - they have all these tools that they provide and then they take a cut of whatever you do. I would say that having tools for people to use and the capabilities for people to use, yes. I'm part of that. The monetization model that was used in the App Store I think is not quite right for this … and most people don't know this, but just so they do know, it's a 30% cut that Apple or Google gets every time people conduct transactions through those. I think that's too high for this market. Because I think what we're talking about is something that's going to be from a volumetric standpoint many multiples of that. So I think there's still a lot of ways for people to make money without charging that. And Amazon's model historically has been to undercut price as they enter new markets, and introducing an App Store model with a very low cost of transaction support on the platform, or no cost. It sounds like something that would be right in character for them and it would really put Apple, in particular, in a bind in terms of how they address it.
Bradley Metrock: [00:51:16] Sure. Absolutely. And it almost makes you wonder ... almost in the way that Amazon responded to Google Home being able to detect different voices and quickly adjusting and integrating that feature ... you wonder if they're waiting to see if somebody else like Google rolls out a marketplace where you can monetize Skills or Google Actions before they do it themselves … it's tough to say. But I appreciate you sharing your perspective on that, and we'll leave that there.
Bradley Metrock: The final story of the week is the only non-Voicebot.ai story, and it's from 9to5Mac, which is an Apple rumor site, primarily discussing the latest post in Apple's Machine Learning Journal. And I include this story because I do hate on Apple quite a bit, because I don't like a lot of what they're doing. But this is not one of those things. I love the fact that they are being more transparent with what they're doing behind the scenes with Siri and their machine learning processes that sort of sit underneath that. And my question for you, Bret, is: Do you share that view? And what did you take away from, not just this post necessarily, but the fact that Apple has a Machine Learning Journal in the first place?
Bret Kinsella: [00:52:43] I think the latter point is the more important one: that they're starting to share. Because a lot of the best minds and most experienced people in this space come from that world, it's an academic world where they publish and they were not going to Apple because they couldn't get the recognition with their peers, they couldn't collaborate with their peers, they couldn't build necessarily off of mutual work that people in the industry were doing. And so I think that was really the key motivator for Apple, so that it would help them from a hiring standpoint and help them with the way that the smartest people in the AI space want to work. I think that's really the more important thing. And the work with Siri is great. I think anybody who hasn't done the study on this in terms of wake words should look at it. It's really fascinating … how you do it, and how Amazon does it as a two-step model local, and then cloud. Siri does a nice job with it. There's some open source things like Snowboy where you can set any wake word, although you have sort of a reliability tradeoff.
Bret Kinsella: One thing I've actually might suggest ... you've interviewed John Kelvie before with Bespoken, and John and I had talked about this very topic … I want to say it was episode 6 when I interviewed him for the VoiceBot Podcast and wanted his perspective because he's done this whole idea of recognition, and trying to understand if that's an initiation of an interaction for some things that he was working on in voice about four or five years ago. And it's really complex. And I think it's great that not only once you get it to work, it's that helpful … but you can see the folks, on the Siri team, on the Alexa team, on the Google's Assistant team continue to invest in that because it's so important that that first point of touch is good. And I find Siri really good in the near field most of the time, but I would say there's still more work to be done there. I would say that the update that Amazon did with Alexa earlier this year, where it does a second verification in the cloud, that that's the wake up … really helped user experience. And I expect a lot more out of this. But this is different than NLU and understanding what people are saying and text-to-speech. It is just the wake up piece which is its own separate challenge and relies not just on the software in the back end, but also the quality of the microphones and the acoustic modeling you're doing at the device level.
Bradley Metrock: [00:55:15] Very cool. Thank you for sharing your insight on that. Thank you for sharing your insights on all of these stories and thank you for sharing your time with us today.
Bret Kinsella: [00:55:23] Well, thank you for having me. I appreciate it, Bradley. You're doing a great service to the industry. I listen every week and I guess this week I won't have to because I was here.
Bradley Metrock: [00:55:33] We appreciate you, Bret. You're doing great work with Voicebot.ai. That is a central resource. I'd just assume that everyone listening to this by now, if they haven't gone to the site yet, they paused the podcast to go there, and also there's a mailing list you can sign up for for a weekly newsletter. I will say this: if there's anyone listening who has a tip for Bret, or has something that they want to share with him or the VoiceBot team, what's the best way to reach you?
Bret Kinsella: [00:56:08] A couple different ways. I'm active on Twitter @bretkinsella. It's probably the fastest way. You can also email me at Bret@Voicebot.ai.
Bradley Metrock: [00:56:17] Cool. Thank you very much, Bret. I appreciate it. For Episode 15 of This Week In Voice. Thank you for listening. And until next time.