Hacker News Reader

Comment by dilap | Hacker News Reader

Having to say "OK, Google" instead of just "Google" (or "Alexa") is a deal-breaker.

Google's products don't feel like they are designed by humans.

>Having to say "OK, Google" instead of just "Google" (or "Alexa") is a deal-breaker.

LOL. That is one of the reasons I hated Alexa. She turned on when a podcast said something about her, the TV ad came on, we talked about our friend Alexa and when anything sounded kinda like Alexa. Ended up changing it to Echo, which worked a bit better. Amazon was out of the question when order to much from there and so common conversation have Amazon in them.

I enjoy Googles way more and no one I know still uses "OK, Google" its much simpler and rolls off the tongue to say "Hey, Google!". First world problems after all.

>Google's products don't feel like they are designed by humans.

And Amazons does? Over half of the stuff I said to Alexa comes back as "I don't know how to help with that" even simple facts. Alexa seems to be very limited in researching and answering. While google has there powerful search engine. Also with google I can build shortcuts for even the most advanced, long winded commands and wrap it into "Hey google, protocol seven" or whatever I want.

The word Google isn't as nice to say as Alexa. I don't have the problems you seem to have with Alexa. It works perfectly.

I wonder if your feelings are related to the Bouba/Kiki effect.

Just googled what that is. It could be!

earlier this you I was on a skype conference from home with a colleague called Alex and a client.

At one point the conversation triggered alexa and i had to apologise and quickly pull the plug

Have you ever been in a room with an Alexa unit when people are trying to talk about it? It's like a comedy routine with the device activating spuriously over and over.

I have, and it's pretty funny. We quickly adapted to just saying "A." or "her".

Probably in a few years this whole debate topic will seem quant, as devices will be much at figuring out when we're addressing them than the primitive idea of a wake-word. :-)

This is not feasible with software updates alone on the Echo Dot or Google Home Mini, as it will require better hardware unless you just want your device to constantly phone home everything.

I think they’re saying they changed to referring to Alexia in conversation to A or Her

I'm responding to "devices will be much [better] at figuring out when we're addressing them than the primitive idea of a wake-word". That kind of on-device context detection may not be feasible with the currently-shipping hardware.

The manufacturers want you to buy new hardware at least every year, so that's not a barrier to new features unless you refuse to upgrade.

I'm sure they could do that detection without local updates in the future (disable the output if it detect a wrong context).

Exactly the same here.

You're assuming that these devices have anything to offer even if they worked.

Not an unreasonable assumption, considering their popularity and the fact that people DO use them.

Considering their enduring popularity in mass-appeal science-fiction constructs (stories, movies etc), I'm pretty sure they do have "something to offer".

I want one sounding like EDI from massefect II and II or Orac from Blakes 7

they do! for me, anyway.

i bought an echo dot on a whim, since it's so cheap, assuming it'd be fun to play with, but not very useful.

to my surprise, it's really handy. setting timers, adding food to the shopping list, and playing music are all super-convenient with alexa. turning lights on and off with her is fun too, although i admit, it was already a fairly easy task to do using a conventional, physical switch. :-)

This should actually be fairly easy even now, simple grammar analysis could discover whether the sentence is directed at Alexa or is simply about Alexa.

But of course its much easier to require "Hey"

I keep the name “Alexa” but you can choose a few (“Computer”, “Echo”, and I think “Amazon”).

The problem happens for any name.

Amazon tried to get around it by naming the product "Echo" while making the activation phrase "Alexa" but that means people don't call it "Echo" in practice.

I enjoy it when mine feels like expressing an opinion in response to the commentary of any Alexa Bliss WWE match.

"Google" is an extremely common word in casual conversation. Any human who tested a device that activated every time you said "Google" would quickly recommend a change to a less likely phrase.

I think the point is that Google shouldn't be in the name at all.

Why, people have been asking Google a lot of questions for a long time. I can imagine the marketing department in Google not loving the idea of another name for Google.

I can see your point. They think that there is enough personality in Google as a thing. That certainly makes the most sense of what I've seen.

However I think they missed on how personal a thing it is to talk to someone. I think they should have viewed 'OK Google' as a personality that's like Google's little brother, that will ask Google for you if that's what you need or else will for example check in your diary for you.

Google doesn't need a personality. Seeing it as technology works perfectly especially when it goes wrong. It's easy to say something like, "Alexa is stupid, she can't understand a simple command", but referring to Google as tech, it is much easier to move on to reforming the command. It is also less invasive for me, personally. I'd rather refer to AI in personal terms.

You are missing the point. Even a random name wouldn't work when people talk about it. Alexa is the perfect example! Google did the right thing. Google home is what people talk about and "Ok Google" is what activities the device.

That's because Google didn't give its assistant a personality. People talk about Alexa because personifying the software is powerful, otherwise they'd be talking about the Echo.

Notice how, even in your post, you say "people talk about Google Home", but the equivalent to Google Home isn't Alexa, it's the Echo Dot.

No it works perfectly well. It worked brilliantly with Siri and had done for years before. People talking about it in the same room is an inconvenience to overcome.

It's possible that Google copied Baidu, which used "Ni hao baidu" [0] is the only thing I can think of.

[0]: https://www.coursera.org/learn/machine-learning-projects/lec... (see the video transcript)

What I'm responding to:

> Having to say "OK, Google" instead of just "Google" (or "Alexa") is a deal-breaker.

"ok google" is a dealbreaker to me, but not because of google. who starts a sentence with "ok"? "hey siri" makes a bit more sense since at least you are interacting with it in a normal way. it's still weird but less robotic than 'ok google', and it's also much better than 'hey apple' for that matter

That is why I say "hey google" instead as easier to say. Our Echo gets a lot of false positives where the Google home does not. Google made the right choice in terms of usability.

I don't think the argument is that it should necessarily be "Google" instead of "OK, Google." It could be anything, but whatever it is, "OK, Google" is a really clunky, robotic, and unnatural sounding phrase. Worse, it thinks it's human and conversational, but it's not, so it actually has a sort of auditory uncanny valley feeling.

It's not good.

I'm not saying they should use "Google." I don't care what they use, as long as it's better than, "OK, Google."

The Assistant now supports "Hey Google" as an option, and it's surprising how much better -feeling it is. It's less clunky, and it seems to motivate more conversational queries.

> The Assistant now supports "Hey Google" as an option

“Now” as in “since the introduction of Google Home”. (On phones prior to Google Home, it only supported “OK Google”.)

I feel the exact opposite. "Alexa" feels like a very clunky attempt to humanise something that isn't human.

My first reaction was "a Russian name?" That's curious...

And "OK, Google" is a really clunky attempt to convince me that I'm having a conversation, when I'm clearly not. It's cringe-worthy, even. It sets off all kinds of, "a programmer nerd made this and thinks this totally abnormal phrase is how real people talk" alarm bells. It's actually embarrassing to say out loud, even if you're alone.

we give names to all sorts of creatures that aren't human -- ships, pets, gods, &c. there's a large part of our brain that is basically hard-coded to interact with things assuming they are people-ish, with names and autonomous behaviors.

alexa and google home are still super-primitive, but as they get more sophisticated, thinking of them as people will seem like the most natural thing in the world.

(i do get where you're coming from, but i think it's a perspective that only makes sense for (1) the very technical and (2) is not long for this world.)

(it's sort of analogous to how originally in internet search engines, you had to put in very precise search terms, and unsophisticated people would just put in some sort of sloppy natural language query. but as the search engines got smarter, the precise stuff got ignored and sloppy was the way to go.)

“Hey Google” is the same number of syllables as Alexa.

Isn't it 'Okay Google'? Either way, I do find Alexa rolls off the tongue a little easier. That may be because I have trouble enunciating verbal transitions.

It's both. It comes down to the keyword identification algorithm of course, not the design by Google. Google's decision choice was not picking a random name that can be identified but instead working in their brand name. My bet is that give it a few years and we will have custom activation words within some constraints.

Is it both? The Google Home I was playing with at my in-laws house a couple days ago definitely only seemed to respond to "OK" (people kept forgetting and using "hey", which didn't work).

Works for me, though I've trained my voice on it, so perhaps that helps. I'm 100% it's supposed to be both.

When you're setting up the Home during the training step it makes you say "Hey Google" and "Okay Google". It's definitely a bit hard to notice considering only a single word changes.

The number of syllables isn't the problem, it's that prefixing every request with "OK," feels extremely unnatural -- almost like you're treating the Google assistant in a sardonic or sarcastic manner.

(Also, "OK" is a very unnatural prefix! Notice how you accidentally replaced it with "hey". I watched a 7 year repeatedly do this a couple of days ago with Google Home, concluding it was broken. Compare w/ Alexa, which kids pick up on instantly.)

Edit: OK, not accidental, apparently "hey" is supposed to work too. Not sure why it wasn't working at my in-laws house! I think "hey" is a better prefix word than "OK" (though still not nearly as nice as no prefix [try always addressing your friends by putting "hey" in front of their name and see how it goes]), so I reduce my criticism by half. :-)

AFAIK, 'Hey Google' as a hotword was supported on the Google Home from the beginning.

Interesting, someone else said that too. It definitely wasn't working for us though... Maybe something to do w/ whether or not it was trained to our voices?

In principle, training to your voice should only be necessary to get access to personal content and settings, but it probably also improves wakeword recognition (and the training trains with both wakewords.)

It's what I started with and didn't even realize okay google was also valid. I wish they'd just let us designate our own, or at least give more choices.

This is what I've wondered. Why not let us designate our own??? Or provide a number of choices. It seems only natural that one day users will be able to name their assistant.

We'll look back at fixed naming and wonder who ever thought that made sense.

If I want my assistant to be named Jarvis, it puzzles me why I can't do that. It would also simplify the case where multiple devices are in the room (e.g. not having to yell towards the specific target device and hope the others ones don't hear.)

Something so simple will deeply personalize the experience for many users.

Personalization would require training the network to recognize your hot word, which needs to work in all sound environments, and detect hot word spoken by different people. That network also needs to be optimized to be small enough to be run constantly by a low-powered CPU in standby mode.

I am not saying that it is not doable, it's just there are real engineering tradeoffs present here.

Personalisation would not give advertising of the company to any guests who witness you using the device ... that's my idea of the primary reason.

Yep, most people don't realize that:

1. Hotword detection runs on a minimal low powered local subsystem

2. The two provided hotwords run networks that have been trained on millions of samples in thousands of different environments.

That thing is very heavily optimized. Alexa and Bixbi get far more false positives and false negatives.

Never thought about that, but makes perfect sense. Thanks for the explanation!!

This. The Google home gets far less false positives and why the wake word makes far more sense. You also do not then get accidently recorded like with the Echo.

Aha. Interesting. That makes sense. Well, that's good to know. At least it means it's coming.

If it's primarily a technical challenge that raises the chances of this eventually becoming reality.

But that's still pretty terrible, they now have to support both 'OK Google' and 'Hey Google' and 'XXX Google' for all other countries.

It's not accidental. That's the activation phrase.

I disagree. Ok Google comes extremely naturally to most people I know. I prefer Google’s personification of an assistant - it’s the brand itself. Not some Alexa or Siri or some gender specific role.

Google has taken an absolute brilliant approach towards voice and their popularity shows. It’s a pleasure to use compared to anything else.

>>Ok Google comes extremely naturally to most people I know.

And it sounds incredibly stupid to anyone else within earshot. I can’t get over how dumb my roommate sounds whenever he uses the thing. My immediate thought when I hear him say “OK Google” is that Google must have told him something and he is responding with “OK” to acknowledge. You know, the way people actually communicate.

I get that it becomes natural after a while but so what? Why start with such an unnatural way to activate it in the first place? Why not “Hey Google”? (I know it is supported in google home)

Not sure what English speaking country, or even US region, you come from but “OK [name], [call to action, often one involving both speaker and listener]” is a thing. And it doesn’t have to be a response to something said previously.

Remember, people can do whatever they want with their language, and it becomes a thing even if you personally aren’t familiar with it.

Exactly, when used that way it involves both the speaker and the listener, as in “OK Bob, let’s go to the park”.

Or, if it is used to pretend a command, it is used as a continuation of an existing conversation, like when you are at the doctor and he or she says “Go ahead and get on the scale” and then “OK, now sit here and open your mouth.”

The way google uses it is definitely not natural.

There is also the pause, after you say “Ok google” and it activates search for you.

Anyway, fanboys will continue to downvote I know but the whole interaction is undoubtedly awkward as fuck.

native speaker of english from USA here. starting a conversation or question with 'ok' is incredibly rare and sounds painfully awkward.

It is better than the echo but it was smart of both Apple and Amazon to call their assistant by a human name. Google is punishing themselves by making their product a function of saying 'Hey [massive global corporation susceptible to PR emergencies], unlock my front door'

There are quite a few people out there who don't speak English. Okay is a common word in many languages since WWII I guess. In many languages there isn't a aspirated H.

> try always addressing your friends by putting "hey" in front of their name and see how it goes

My friends are better at contextually disregarding comments and don't send any utterance that seems to invoke them to a remote server, so don't need a form of address designed to avoid false positives.

Also the same number as "Hey, Siri". And Samsung's "Hey, Bixby". I'm assuming it's not a coincidence.

I think it's likely that three syllables just happens to be a nice cut-off point that minimizes the number of spurious activations, and that explains why everybody doing voice control uses that particular threshold to activate their digital assistant.

Heck, even Star Trek got it right way back in 1966 with its "Computer" wake word.

About the only company I can find that seems to be ignoring the three-syllable rule is Microsoft, who use "Hey, Cortana", which is four syllables. But Microsoft's always been a fan of using really long phrases to identify their products, so maybe that shouldn't be a surprise.

Just be glad that VMware don't make one.

"Hey, VMware vSphere Home Voice Assistant Essentials Plus..."

look how many times the family has to remind their grandma to say "ok" or "hey" in front of google in this video:

https://www.youtube.com/watch?v=e2R0NSKtVA0&feature=youtu.be

not designed for humans! i rest my case.

The thought of millions of children growing up in "Google homes" being programmed to yell out a brand name constantly, is incredibly sad.

Reply on news.ycombinator.com