Sometimes you see something new-fangled & you pause and say “what?” That’s how I felt about Amazon’s “Echo Look.” To be fair, I read about it before I saw one in action. And I’m a typical guy who doesn’t care how he dresses much (a lot of t-shirts). But after seeing this video showing how it works, it piqued my interest.
The “Echo Look” will take pictures (or a short video) of you trying on an outfit using a voice-activated camera (“Alexa, how do I look?”). It will brighten the pic & blur the background so the picture looks good. You can then perform a “Style Check,” which uses machine learning comparing two of your outfits (based on the pictures you just took) against current style trends & what flatters you.
The device also can create a “Personal Look Book” to help you keep track of your favorite outfits. Crazy for a guy like me to do so, but I had to get one. Well almost. I read the reviews and you can’t crop the photos nor do the photos get saved to your phone’s “photo gallery” automatically. But I was tempted because comments said they took very nice selfies. I’m the self-proclaimed “Selfie King.”
And I don’t care about fashion advice – but if I did, it sounds like that feature doesn’t work that well if you believe this comment:
Let me put it this way. When you are buying a high-end product with some huge AWS AI neural network behind it you expect it to be something more than a selfie-taker! I want conversations like:
– Alexa, how am I looking?
– I think that black skirt you were wearing last Sunday would match this top better
– Alexa, am I underdressed for a friends wedding?
– I think high heels would suit better than flip-flops for this occasion.
I want it to have some sort of intelligence, not just randomly assigned percentage score for each look.
So when I figured out that “Yes Sire” wasn’t appropriate for my 9-year old nephew, we tried the “Magic Door.” It was a good choice. It was an adventure game with a magical land, where you collect hidden items, solve riddles and help creatures. Tolkien stuff. “Alexa, open the magic door.”
As Scot Westwater discusses in his “Pragmatic Talk” podcast (episode 5), when building a skill, you’re designing a conversation. Having a conversation with a voice assistant should feel like you’re talking to a human; not talking to a robot. So it shouldn’t feel like a stilted FAQ-style back-and-forth.
For conversational design, you need to understand that when humans start down the path of a conversation, we tend to “fragment” our statements. Our brains then automatically connect the dots to fill the gaps. The key is to train our voice assistants to do the same – build the logic of how a human would naturally communicate about a topic into your skill.
According to Scot, ask these four questions when building your skill:
1. What words will be used, including possible synonyms?
2. What questions will be asked?
3. How will the questions be asked?
4. What additional utterances might be made?
This is all much harder than it seems – but luckily, you can study the data about how actual users are interacting with your skill & continuously improve it in response. In addition, as mentioned at the end of episode 7 of their podcast, the Westwaters discuss intelligence gathering to uncover how users are discussing you – the obvious place being social media (Twitter, Facebook, LinkedIn, etc.). The comments provided by your customers there can be useful to understanding how they think & what they might say to your skill…
When marketing your skill to your customers, don’t tell them just “here’s our new skill” – explain exactly how it will make their lives easier. Tell them here’s the valuable thing we have given you. But then you need to really get the word out through your traditional marketing channels. It’s not a “build it & they will come thing” – it’s early days in voice, so marketing truly is key…
Some of the voice products available seem like a stretch to me. One of those is the “Echo Wall Clock.” At first glance, it seems fairly cheap at $30 – but you can buy a cooking timer for only $3. Of course, you don’t need to even spend the measly $3 if you already have an Echo – because your voice assistant can serve as a timer for you. But using your Echo as your timer does have a disadvantage – it can’t display how much time you have left. Hence, Amazon is selling this wall clock. “Alexa, set a 12-minute timer” – and see time count down at a glance.
Note that the Echo Wall Clock must be paired to – and within 30 feet of – a compatible Echo device. So the $30 becomes $60 if you don’t already own a voice assistant. And one other limitation – when your compatible Echo is playing music in stereo pairing or speaker groups, or multi-room music, your clock will be temporarily disabled. Since many people play music while they cook, this may be a problem for more people than it might seem. And finally, some of the clock’s reviews indicate it is shoddily made – the hour/minute hand apparently is fragile…
Recently, I blogged about this free 48-page playbook by “360i” about what you should know about voice from a marketing perspective. The chapter that begins on page 48 explains how to apply SEO principles to voice. It includes a diagram that maps out how user intent pathways – and the natural language used – reveal preferred content sources for Google Assistant.
Natural language inputs and the order of words in a request give us clues to the true intent behind a user’s discovery outcome. While keyword searches do this too, longer voice-driven queries add linguistic details that can be used by assistants leveraging machine learning to determine what content type or delivery is most appropriate to the moment and situation.
And this excerpt drives home the point about how important SEO is for voice:
With more clues towards consumer intent in Voice, the same search algorithms will have more data to source from. When searching or querying, consumers use more words when speaking than they do with typing. Typing keywords into a search bar is a sort of truncated version of natural language. As competition to be the top (and now sole) answer, mapping consumer intent to content experience will play a much bigger role in SEO.
I’ve been wondering about the problem of interacting with a voice assistant in a room that has a lot of people being loud. Wondering how do you solve the ambient noise problem. This article blew me away because it describes a project where a voice assistant would essentially be able to read lips. Here’s an excerpt:
There are lots of digital assistant commands one should never say out loud in front of other people. But in a few years, proximity to eavesdroppers might not be a problem anymore, because we’ll be able to silently mouth all that embarrassing dictation. Called “silent speech recognition,” the technology essentially reads lips. Users mouth the words of a command or message, and the device deciphers the words, based on the movements of the face and neck.
The feat can be accomplished using surface electromyography, or sEMG. Electrodes are placed on the skin at key places around the mouth, along the jaw, under the chin or on the neck. Muscle movements in those areas generate neuromuscular signals—essentially an electrical code. Algorithms trained on silent speech then translate the electrical signals, decoding what the user is saying.
As the article notes, this also could solve the problem of allowing you to not say confidential information out loud – for example, a password – in a place that is not private…
When I talk to people about the potential of voice, a common reaction is that they aren’t interested. Note that most of these people are around my age – over 50. Even when I explain the myriad of ways that voice can make life better for so many, I get shrugs. One of the biggest concerns is privacy. That’s why this note from the “RAIN” agency is welcome:
In response to multiple press reports and challenges from regulatory bodies, the tech industry has taken steps to reduce human oversight of voice recordings and provide users with more control. For now, Google and Apple have ceased human reviews of voice data transmitted through Assistant and Siri respectively. While Amazon has not halted their processes, they have created an option in the Alexa Privacy Settings for users to opt-out of their data being reviewed in this way. It seems the latest wave of pressure around data management has been heard by these companies loud and clear.
Voicebot.ai’s Bret Kinsella has this excellent exclusive about how Amazon is now highlighting ad products for brands to promote their Alexa skills. Amazon’s tagline for this is a hilarious take on the famous line from “Field of Dreams”: “If you *only* build it…they won’t come.”
Here’s the intro to Bret’s piece:
Amazon forbids ads within most Alexa skills, but it is now highlighting its own ad products that enable marketers to promote their skills through Amazon’s advertising channels. New materials distributed today to marketing agencies include familiar information about identifying Alexa skill opportunities, choosing invocation names, and designing successful voice solutions. There is also a section called “Alexa Marketing & Advertising Guide,” that begins with the statement, “if you *only* build it…they won’t come” [emphasis original].
Loved this interview by Bret Kinsella of “voicebot.ai” with Tellables’ Amy Stapleton. Amy is one of the first people to use voice assistants to tell stories, as she uses her company, Tellables, as a publishing platform for conversational stories. Here’s some of the cool things that I learned during the podcast:
1. Amy distinguishes how her “conversational storytelling” platform differs from “games” even though someone using it gets rewarded in some ways. Her platform offers storytelling content with an interactive component. But what Tellables does isn’t quite gaming even though some “choose your own adventure” games have some storytelling in them.
2. Amy’s “Tricky Genie” was one of the first stories available on Alexa, enabling her to gain significant rewards through Amazon’s reward program (being a first-mover was important; Amy believes it’s important to build an audience first before trying monetization). “Tricky Genie” is a one-on-one experience with over 100 scenarios available.
3. Amy’s latest offering – “My Box of Chocolates” – uses a “Polly voice” to tell a short story. The Polly voice selected for a particular story has a personality that fits that particular story. The stories – typically 200 words or less – can be heard by yourself or in a group. Her goal is to offer stories that make you think. The interactive components at the end of the stories help to get you thinking. The interaction hook is an important way to make the voice experience special. After enabling the skill, “Alexa, open my box of chocolates”
4. At the end of the short story, a “party question” is provided. If you’re convened as a group, the party question is a great way to provoke a conversation. So you could hold a book club meeting and listen to “My Box of Chocolates” as a way to mix things up for a change. And saves folks the embarrassment of saying they haven’t read the book!
5. So in a sense, this type of platform is the flip-side of the danger of screens taking us further & further out of our communities – with voice, there is an opportunity to bring people back as a community.
6. For “My Box of Chocolates,” Amy reaches out to authors to submit short stories. Since stories being told by a Polly voice on a device is different than reading short stories, there is a bit of an art to creating content that works on this platform. So Amy winds up doing a little bit of training for authors that are new to this.
7. When you build a skill, Amy recommends that you build it so that you can continuously add new content. Keep people coming back from more.