Apple’s Siri voice assistant has made consumers enthusiastic about using a Voice User Interface (VUI) to interact with their mobile phone. Siri has created a good impression on the user with natural language processing which makes it possible for the user to say ‘anything,’ as if they are talking to another human. Siri also has a ‘personality’ and sense of humor. Apple has done a good job marketing Siri as a personal, easy to use voice assistant, but Siri is not the first voice assistant on a mobile device. Google has had a voice assistant application called ‘Voice Actions’ that runs on the Android platform since August 2010.
About Voice Actions
Voice Actions allows the user to say specific commands to get things done such as sending text messages, sending email and listening to music. However, Voice Actions lags behind Siri in several areas. For example:
- Voice Actions does not have the Artificial Intelligence capabilities that Siri has to provide ‘intelligent’ answers to questions that are not related to the commands it knows.
- Unlike Siri, Voice Actions has no Text To Speech (TTS) capabilities to ‘speak’ back to the user.
- Voice Actions is also unable to have a multi-step ‘conversation’ with the user like Siri.
In response to Siri, Google is planning to make major enhancements to its Voice Actions application. This effort is called Majel.
About the Majel Theory
Majel comes from Majel Barrett-Roddenberry, better known as the voice of the Federation Computer from Star Trek. The approach that Google is supposedly taking for Majel is to make everything on your phone voice aware. It is like the Starship Enterprise from Star Trek where you can use a VUI to interact with practically anything on the ship. Majel is not attempting to give the user the illusion that it is interacting with a ‘human’ but instead it will be clear that you are interacting with an intelligent computer. This is in contrast to Siri, where Apple has put in a lot of effort to give a ‘personality’ and a sense of humor to the responses you get from Siri. Click here to read more »
This year saw the debut of the first wide-scale voice user interface, Apple’s Siri. While there have been a lot of articles making fun of it, it marks a shift in how we interface with computers, bringing us ever closer to the model laid out in Star Trek in the ’70s of simply talking to a computer. In the short term, the greatest impact will likely be in how consumers view speech technology interactions. The most prevalent speech technologies are IVRs, from which people are going to demand a more natural, intuitive experience and the industry is going to have to respond.
But how are they going to do that? What are the keys to a good caller interaction, and how do you implement them? The first aspect to explore is: what’s the secret to Siri’s success?
Is it the natural language interface? Yes and no. While many people enjoy the ability to ask, “Do I need a coat today?” and have Siri answer with a “Yes, the temperature will be 35 degrees,” that is not the real key to its success. Additionally, this is not a realistic solution to most large call center situations because callers often don’t know exactly what they want when they call a company and tuning a natural language interface can take years and millions of dollars to complete.
Is it the deep ties into the user’s personal data? Yes. This is good news for any company handling a large call volume from a diverse set of customers, because one thing companies have is a lot of data on their customers.
So what does this mean for the industry? It means that companies will need to start using all of the available caller data to create intuitive, proactive interfaces that can make educated guesses as to why a caller is calling and about how to best handle them and their individual needs. When I log into a website, the Web page immediately shows me my profile, preferences, and frequent interaction types; why can’t a phone interface do this as well? The first step will be to link CRM systems and customer profiles to IVRs, and the second will be to build dynamic systems that can respond to different callers personally and to diverse caller situations in a more efficient manner.
Click here to read more »
I was at a carnival night at my oldest son’s elementary school last week and while standing around making small talk, one of the parents proffered that age-old question (that I often dread!): “So, Kit, what do you do?” Since I work in the IVR industry, I had to fight back the tendency to say something smarmy such as magician or astronaut, partly because it’s more fun to be one of those things and partly, I must confess, because I am a little bit ashamed of my industry’s reputation. I know, sacrilege, right? But I said it out loud and as I did, I realized something—this must be what Harry Potter feels like when he says Voldemort and makes everyone around him cringe. Very liberating indeed!
Now that I have not only pointed out the white elephant in the room, but have shined a floodlight on it, we can move on to the more important issue. How do we fix it? I learned a long time ago that it is extremely easy to point out the problems and flaws in any situation. The real challenge isn’t in being critical, but rather in taking the road less traveled and finding a solution. And while I don’t have a magic bullet, per se, what I do have are three cardinal principles that I apply when talking to customers about self-service. One is mine, one is borrowed, and one is flat-out stolen, but I make no apologies. After all, imitation is the sincerest form of flattery!
Is it Better Than Human?
This should be obvious, but I find that, more often than not, it isn’t. It seems many companies apply the “It’s better than nothing!” theory. With every project you consider, just apply this test: is it better than a human experience? Or, put another way, is it better than waiting on a human? When I call my airline, I am thankful that it knows who I am and that it tells me immediately my flight status, getting me off the phone in under 30 seconds! There is no way I could get through to an agent, even with no hold, provide them with the necessary information, and get my flight info faster. That is a perfect example of better than human. On the other hand, the other day I called my bank to transfer funds from my account to an external account. The system tried dutifully to guide me through the process, and I finally completed the transaction (10 minutes later!). Oh, how I longed for a rotary phone I could slam when I was finished! Really!
Click here to read more »
Apple’s Siri on the iPhone 4S—just released in October—has already been welcomed by the media as both an epic fail and as groundbreaking at the same time, which is not a huge surprise for a new Apple product (think iPad sans native Flash player). Rather than fuel the fire in this debate, I would like to focus on how remarkable it is that what Apple has achieved to do is bring to the forefront a very visible implementation of speech-enabled IVR—no, multimodal—technology with takeaways for any business looking to incorporate these technologies to better serve their customers.
One thing that makes Siri interesting is that it utilizes natural language processing. But really, in its current state, it does:
Do a few things well, instead of trying to perfect every possible outcome.
This sounds simple enough, but for businesses looking to design a voice solution with a high level of self-service, it is important to callers that an IVR perform certain automated tasks (such as looking up account status or making a payment) efficiently and effectively, instead of performing five other functions that are complex in nature or are error prone.
In Siri’s case, when it is first activated, you can see a list of accepted commands by pressing the “i” button, or by simply saying, “help.” So Siri is pretty clear on the types of things it should do well. However, I’ve found something else at which it is good that people are generally not: math—well, at least two-operand math: Do you ever wonder what the per-unit cost of the 100-piece bag of Halloween candy is versus the 40-piece bag? Or what is really 30% off of $12? Just ask Siri. Or I suppose you could crunch the numbers in the Calculator app, but the takeaway here is to:
Give your customers choice.
While the Internet has wreaked havoc on customer service centers in the past 15 years—with the introduction of email, Web forms, live chat, SMS, Facebook, Twitter, and so on—the real story is about giving customers the option to choose how they want to interact with you and how they would like to be served. Click here to read more »
« Older Posts
One of the selling points of Android is that it is available on a large number of smart phones and tablets that have different form-factors and it is now even on Google TV. This widespread use is an advantage in one way, as Android applications are then available on a large number of devices, but it is also a challenge, as applications designed for Android will need to look good on devices with different screen sizes and dimensions.
There are two primary methods for making an Android application’s User Interface scale to fit different screen sizes and dimensions.
The First Method: Getting the OS to Scale
The first is the simple approach: design a single layout that the Android Operating System (OS) can scale to fit different screen dimensions. There are several rules to keep in mind when designing your User Interface if you take this approach.
1) Specify rules for the size and location of widgets on the screen instead of specifying exact locations and sizes for the widgets. The Android OS will determine the exact location and size of the widgets based on the actual screen size.
2) Instead of using pixels to specify the size of widgets, consider using millimeters (mm), inches (in) or density-independent pixels (dip) as appropriate. With pixels, you cannot control how big a widget looks on the screen. If the density of the screen is low, then the widget looks big—but if the density of the screen is high, the widget looks small, as the same number of pixels are confined to a smaller area on the screen.
3) Use scalable nine-patch bitmaps and XML-defined drawables instead of using classic bitmaps (i.e., PNG, JPG, GIF). Classic bitmaps will be scaled using normal scaling algorithms by the Android OS and may not produce a good result. Nine-patch bitmaps are PNG files that have encoded information on how the image can be scaled. XML-defined drawables use XML to define the shape, strokes and fills, so they can be scaled with no loss of accuracy.
Click here to read more »