Tellme on Windows Phone vs Siri on iPhone

In a Forbes interview with Microsoft’s Craig Mundie, Chief Research and Strategy Officer, he contended that comparing Siri on the iPhone 4S and Tellme on Windows Phones are much alike. Craig says,

At least from a technological capability you could argue Microsoft has had a similar capability in Windows Phone for more than a year… All that is already there, fully functionality for years.

I respectfully disagree. Tellme might be state of the art speech recognition technology, one of the most accurate and adapted to different accents, however its commands are still keywords syntax-based. “Call John. Find pizza.”

Where scenarios exceed the confines of syntax-based functionality, Tellme doesn’t have much to offer. The example Craig used of messaging, “message Person, say message, send” is more of an exception than the rule.

In comparison, I regard Siri as a speech interaction technology. Because it places much more emphasis on context, it appears to be accurate enough for the limited set of functionality it provides. Speech recognition might not be new technology, but Siri definitely puts a pretty face on it. It remembers your name, responds through voice and makes jokes.

“What’s the weather going to be like this weekend” is what you expect in a conversation, not “Find weather in Melbourne” which is like computer talk.

Now to say Apple invented this would be false, Apple bought Siri – a very smart acquisition. Of course Tellme, which was also acquired, has plans to make itself more natural too and they most likely will, but that’s not happening overnight. Siri is available today.

The next big phase of Tellme is in the upcoming XBOX dashboard update where it will power Bing search. You could say “XBOX Bing Batman” but the XBOX doesn’t talk back, like “Would you like to watch The Dark Knight?”. Although it makes you wonder what a voice for XBOX might sound like.

16 insightful thoughts

  1. On the phone, I really wish it’d give me a chance to change what it recognized. For example:

    I say: Call Lina’s mobile
    Recognized: Call Linda’s mobile [calling in 10 seconds...]

    but each word underlined and give me smart-tag / auto-correct ways to change it to something more relevant. So I can tap on the word Linda and change it to Lina.
    and learn my voice over time dammit.

  2. Yes, Mundie is either trying to put on a brave face or just completely clueless. Siri pwns TellMe on the NUI side, which is what he’s always touting as a core focus for MS.

  3. Yeah got to agree with you there Long. Apple managed to humanise Siri with the commedic responses, while being a lot more functional with regards to setting alarms, meetings, reminders etc. WP7 voice functionality still feels like last generation voice commands, regardless of how Microsoft try and sell it.

    The problem isn’t with marketing as the video suggests, but rather the lack of a response to Siri. TellMe was probably on-par with Apple’s voice controls like Play Artist xyz in iOS4, but with Siri, Apple takes it to the next level. Microsoft needs to respond, release an update and not only match the functionality of Siri, but surpass it to draw attention and be a point of difference for potential smart phone buyers I look forward to that day.

    Also Siri its worth keeping in mind that Siri is still in beta. While we are yet to see an update, when it comes, it’s likely to add even more voice controlled functionality. The way I see it, there is a race on to be the first to have an OS-wide voice command API that applications can integrate with. When that happens, it will be a game changer.

  4. I think you perhaps over-estimate the importance of “humanizing” voice search on a mobile device. As long as people get the information they desire using only their voice I doubt they care whether or not it is keyword based vs. having an awkward conversation with your phone. In fact, I think the point you make about the TellMe stuff being more well adapted to a much wider audience is a much more important facet to speech (and speech adoption) in the current generation of technology than having some AI that “just gets you”.

    1. I agree with you people probably don’t care whether or not it is keyword based vs. having a conversation, but from a marketing and usability perspective, you’ll probably find more people will use something that is easier to discover and is a lot more flexible.

      For both systems, you can tell it to fetch the weather information. On Tellme, I knew such feature existed but I didn’t know whether the right syntax was “Melbourne Weather” or “Weather in Melbourne”. Trying “Melbourne Weather”, it did something unexpected, it started my “Weather” app. Although that also fits my purpose, it wasn’t what I wanted.

      With Siri, I don’t know what the “right” syntax for weather is either, but I can guess by asking if I was asking another person. Of course I’m sure I can also be short and concise phrases like “Melbourne weather” to do the same thing.

      Yes it’s true Tellme might work for more people (more languages, more accents), but as I’m sure the last decade has illustrated, user experience rightfully trumps features and functionality.

  5. Don, I think you nailed down the point. I couldn’t be bothered saying complex human sentences to a machine. Just short and concise is the way to go.

  6. I agree with Don, too. Yes, Siri is different but most of the time there are only needless words added to make the input and output sound more human. As for your example “What’s the weather going to be like this weekend” the only relevant information is “weather, this weekend.” Sometimes Siri is a bit too chatty for my liking.

    Just for the fun: Imagine the next Xbox dashboard would allow you to say: “Oooh dear Xbox, would you be so kind and pause this video now for a second?” ;)

  7. I support Long’s arguments. I have voice control in my mobile and I have voice control in my car. I end up only using the voice control in my car, because otherwise I would mix up the syntax of both systems resulting in an trial and error approach to voice control. So as Long stated: If you don’t have to remeber syntax voice control get’s a lot more intuituve and ubiquitous.

  8. Siri is also completely keyword based, it just ignores words it doesn’t recognize as commands or modifiers of those commands. If TellMe got some simple filtering added it could do almost exactly the same things Siri does.

    1. That’s great. So why didn’t they do this before Siri/Apple? You can’t tell me they’re not smart enough to understand that this would be a desirable feature, or indeed how to do it. They’ve got smart people there. So why wait for a competitor to show them up, then have to plan an upgrade/response? Why not just make it better?

Leave a Reply