March 8, 2016

Hound is out of Beta, Voice Assistant Apps Take Advice of a Toddler

By Rina M.

Hound just escaped from its beta stage and is about to take this voice assistance thing to a whole new level. Before you begin to make witty, passive jests about yet another shoddy voice assistant app, let’s dig deeper into this newly debuted platform.

Accompanying Google Now, Apple’s Siri, and Microsoft’s Cortana, Hound declared that they might be a wee bit more awesome than its counterparts.


Hound was developed from the SoundHound platform. Aiming to cripple Google Now, Hound is supposed to be phenomenally faster and more precise than Google’s assistant in terms of utilizing speech-to-meaning recognition, which uses the same developing operations as SoundHound. SoundHound’s President and CEO, Keyvan Mohajer, detailed that Siri, Cortana, and OK Google differ from Hound by using speech-to-text and text-to-meaning: “If you make a mistake in speech-to-text, you’ll have the wrong input in the next step.”

Furthermore, Hound is supposed to comprehend more intricate questions, eliminating simply phrased questions.

So, I suppose asking Hound a query like this might suffice for a coherent answer in return?
“Hound, I’m in a bit of a pickle and can’t elect between Tex-Mex and sushi to satisfy my voracious appetite. Can you tell me the nearest four-star rated sushi bar on Yelp located less than 10 miles away from my house, besides Maki Yaki?”

Highly improbable. But that’s what we’re here to find out!


Hound vs. Google Now

Sure, Yelp and Uber have joined Hound’s party, integrating their APIs and providing the user the ability to request an Uber ride, receive Uber price estimates, and efficiently use Yelp’s directory for local searching all through Hound. However, the skeptic in me wanted to challenge both Google Now and Hound to see which voice assistant can serve more in my favor.

When I asked Hound and Google Now how long it’ll take me to get home at 3:30pm today, both responded but with very different answers.

On the other hand, Hound did too much, too soon. Granted, their query input was precise. However, to answer my question, Hound fetched me a YouTube video to a Phil Collins song and a website about mortgage approvals.  I mean, sure, we can use some Phil Collins every now and again to serenade us to slumber, and I’m sure a mortgage approval will come in handy someday, but Hound!!  You didn’t tell me what I needed. You had one job!

Later, I asked Google Now and Hound if it was going to rain tonight in my city. Both certainly returned a reply, yet Hound said it was going to rain and Google Now stated that it was not. This opens more doors for ambiguities since both platforms may (by default) be using different weather applications.

I will now walk home weather-conflicted, with a sweater tied to my hip and an umbrella ready to employ as my armor.

The verdict: After just a few searches, Hound seems more organically receptive grasping by request faster compared to Google. Yet Google Now, seems to grasp my needs more contextually and, thus, better than Hound. 

I know what you’re thinking: throw a dog a bone. Sorry, Hound, this thing that we have here, it isn’t going to work. *Uninstalled*

It’s a no brainer that Hound, Siri, Google Now, and Cortana all have a long way to go. Essentially, all electronic gadgets and machines will always have room for development. I mean, we’ve all heard about Google’s self-driving car and its recent, minor calamity.

You’re Not Smarter than a 3-Year-Old, Google.

Hector Ouilhet, design lead for all Google’s search products, looks to his three-year-old daughter, Anna Julia, for some of his inspiration.  He discovered that users need to continually see Google Now and all other voice assistant apps like a three-year-old child.

Google and the rest of us egotistical adults aren’t as smart as Anna Julia.  Ouilhet strives to train Google Now to respond more effectively to user requests, creating a voice assistant model that is intelligent, capable, evolving, and have feelings, dang it – just like Anna Julia! 

To further illustrate this point, Ouilhet wants us to realize that, just like Anna Julia, these voice assistant apps can evolve with age, when programmed more efficiently from observing and learning a human’s behavior and requests.

Anna Julia may not respond semantically to her father’s request. Say he proposes that she set the table for dinner. To address her father’s request, she’s going to try, even though she doesn’t understand exactly what “set the table” actually means. Anna Julia may not ultimately succeed in what her father asks, but she will…
1. Respond in any sensible way that addresses his request from what she empirically observes.
2. Respond the best way possible that she’s capable of understanding to her capacity in order to fulfill his requests at all.

She does this by atomizing his statement into intricate parts. All in all, she may understand the context, but the results won’t be precise.

Sounds familiar? Well, it hits close to home whenever I ask Google to text so-and-so “Hey, It’s Rina. Call me later,” and they translate my name to “Arena,” compromising the entirety of the text message. Who’s Arena? AM I NOW IN THE HUNGER GAMES? It’s not fair, I tell you. 

Human vs. Machine

Let’s understand the nature of what we are dealing with, here. Sure, voice assistants speak like humans, but they aren’t like us. Computers suck sometimes.   Google now isn’t going to punctuate our sentences all the time. It’s going to spell names inaccurately, and it just might reduce to Googling our request because it isn’t   designed to articulate a voice-response for us at that given moment.

The more humans and our robots evolve, the more we have yet to accomplish. Technology is doing more than just inputting a query and fetching an answer. They are perfecting semantics and honing in on context. It’s no longer about what we are searching, but understanding the why, when, where, and how. It’s a relationship, still buffering, between humans and machines.

Do you think it’s possible that machines, AIs, and robots can be just as smart as humanity or even smarter in the distant future? Haven’t you seen Ex Machina?!Will our best companions be engineered remarkably out of metal and a complex motherboard? What are your thoughts? Fears? Hopes and dreams? Let the Tekkies know!

No comments:

Post a Comment