Hacker News Reader

Comment by josteink | Hacker News Reader

Ok. So 100% voice-based. That makes sense. That's great. Except if you don't speak English as your primary language.

I've tried Google's voice-recognition on my native language and the results have been closer to sad than lulz-worthy.

Yes, I know English. Yes, I can both write and speak it fairly well. That's not the point: You won't believe how awkward a context-switch from "I'm talking to people" to "I'm talking to this machine, but I need to do so in English" actually feels.

It's just feels so fundamentally wrong, that despite how it might save you time and effort you always go for the option where voice-recognition is not involved.

Unless Google improves their voice-recognition for everything not US English, I predict this product failing horribly in most markets, even with everything else 100% perfect and flawlessly executed.

View on HN

This seems like a silly beef to have with a revolutionary product that still is in beta. You know it's developed by an American company, right? It's not crazy for them to have the first language supported to be the language the people from the company speak. Give it some time.

Voice recognition has always been patchy or non-existent for languages other than English, and it's a deal breaker for something that relies _primarily_ on voice for input. That's the point I believe he was making.

I don't think he was referring to the glasses in specific rather than the state of voice recognition as a whole.

The voice recognition has been in android for 4 major versions now. It no longer hold the beta-badge protection and needs to start showing some real world results to be taken seriously.

The Google Glass device probably uses the same voice recognition as Android. Android is not in beta anymore, yet its voice recognition fails pretty bad on anything other than English. Google Now is almost entirely useless unless your device is set to US English.

I get that they can't exactly make everything work perfectly for everyone. But most of the planet is not american. They want to interact with technology on their own terms, in their own language. Google is long past the point at which it can make excuses now. While breaking new grounds with stuff like Google Glass is very cool, so is offering useful voice recognition to most of the planet.

The big difference between the two is that Glass is powered 100% by voice, Android is powered primarily by touch. I am sure that by the time Google starts rolling this out to the mass market in other countries, they will have those problems figured out.

Yeah, it's too bad that Google will never think to improve their voice-recognition software beyond English.

From a demo I tried, the computer on your right temple is at least sensitive to swipes and taps to access a menu. In addition, the menu can be scrolled by looking up and down, so voice is not necessary.

For what it's worth, Google Now works pretty well in French, so I'm confident they'll manage to do something usable after some time. Of course it won't be their first priority...

I would always choose the non-voice option if possible, even in english, but can anyone think of a useful way this could work without it?

eye gestures?

Maybe they could link to a phone/tablet or a pebble style watch, alternatively I remember seeing Microsoft have done some cool by projecting a touch interface onto skin. Perhaps the user could do gestures which are picked up by the front camera but it may make everyone look like they're being attacked by invisible bees all the time.

Google Glasses remind me a bit of the time I tried to work out how Tony Stark actually controls all of the functions of his Iron Man armour, and what sort of (realistic) system might come close to what's shown in the movies. I didn't get very far.

To me, Glasses are an awesome concept that begins to fall apart as soon as I think about them - I think they'll either need to pull a major and completely surprising leap forward in how we interact with tech, or at some stage we'll need a keyboard to use them effectively.

Voice-based input also has massive potential for trolling. I'm not keen on the idea of having a stranger walking next to me text my family through my glasses.

I'm thinking that once the promo videos are over and forgotten, the majority of the user input will be through a touch device linked by bluetooth. It just makes more sense to me from a practical standpoint.

Since it's attached to your head, it may be possible for them to match vibrations with speech input and verify that you're the one who said it. Would help in noisy environments, as well.

That's a really good solution - from memory a similar system is used on the headsets worn on Sydney Harbour Bridge tours. I'd forgotten about them until you mentioned vibrations.

I'm actually going to be disappointed if they don't have this in some form. I guess the most elegant solution would be to read skull vibrations rather than use voice at all - I'm not sure how viable that would be though.

That's not for voice recognition, it's to replace the headphone SPEAKERS because it's so windy on the bridge. The thing sits on the bone just in front of your ear. It's a really funny sensation but works well.

But I assume the same system would work in reverse.

I know that the bridge headset replaces the speakers, but reversing the system would remove a lot of the inherent problems with a system that relies on speech for input.

I remember using them when I was on the Harbour Bridge what feels at least 5 or 6 years ago - so maybe the tech has developed enough since then. Especially if Google has been throwing resources at it.

The voice dictation works through bone conduction.

This makes me happy - I remember the headsets I've used that used this in the past needed a lot of customisability to get a fit flush against my head. It was about 8 years ago though, so I'm assuming things have come along since then.

Do you have any idea if sound will be made though bone vibration or traditional speakers or both? I'm still a bit skeptical about how this is going to work in crowds.

It seems like audio is transmitted solely through bone vibration based on the FCC report https://apps.fcc.gov/eas/GetApplicationAttachment.html?id=18...

Not too mention all the trouble when you live you life in two different languages... somehow, none of all language aware things are very practical when you communicate sometime in english, sometime not.

The solution is simple: always speak English. This fixes other similar problems.

The other problem with their voice recognition is speaking English with an accent that isn't US or RP.

I speak English and their voice recognition is laughable for me too.

Reply on news.ycombinator.com