Google Pixel Buds Are Wireless Earbuds That Translate Conversations In Real Time (arstechnica.com) 163
At its hardware event today, Google debuted new wireless earbuds, dubbed "Pixel Buds." These are Google's first wireless earbuds that give users access to Google Translate so they can have conversations with people who speak a different language. Ars Technica reports: Unlike Apple's AirPods, the Pixel Buds have a wire connecting the two earpieces. However, that wire doesn't connect to a smartphone or other device. Pixel Buds will pair via Bluetooth to the new Pixel smartphones -- and presumably any other devices that accept Bluetooth wireless earbuds. All of the Pixel Buds' controls are built in to the right earpiece, which is a common hardware solution on wireless earbuds. You can access Google Assistant by tapping or pressing on the right earbud, and the Assistant will be able to read notifications and messages to you through the Buds.
But the most intriguing feature of the Pixel Buds is the integrated Google Translate feature. Demoed on stage at Google's event today, this feature lets two Pixel Bud wearers chat in their native languages by translating conversations in real time. In the demo, a native English speaker and a native Swedish speaker had a conversation with each other, both using their native languages. Google Translate translated the languages for each user. There was barely any lag time in between the speaker saying a phrase and the Buds' hearing those words and translating them into the appropriate language. The Pixel Buds will use Google Translate to comprehend conversations in 40 different languages. Some other features include a 5-hour battery life, and a charging case that can hold up to 24 hours of battery life. They're available for preorder today for $159.
But the most intriguing feature of the Pixel Buds is the integrated Google Translate feature. Demoed on stage at Google's event today, this feature lets two Pixel Bud wearers chat in their native languages by translating conversations in real time. In the demo, a native English speaker and a native Swedish speaker had a conversation with each other, both using their native languages. Google Translate translated the languages for each user. There was barely any lag time in between the speaker saying a phrase and the Buds' hearing those words and translating them into the appropriate language. The Pixel Buds will use Google Translate to comprehend conversations in 40 different languages. Some other features include a 5-hour battery life, and a charging case that can hold up to 24 hours of battery life. They're available for preorder today for $159.
local processing (Score:5, Insightful)
We all know that processing is not done on the phone.
Which means the real headline should read "Google earbuds will send every word back to the mothership for processing."
Re: (Score:2)
We all know that processing is not done on the phone.
I'm not so sure about that. There's lots of fairly advanced stuff that easily runs on mobile devices, like speech-to-text. The days of "everything runs in the cloud" are over.
Re: (Score:2)
Unless something has changed recently all that speech to text is not done on the phone, its all offloaded to servers. There are companies out there that most companies like apple, google, amazon, all the tv providers, etc, contract out to to handle their speech to text needs. They return back the text along with lists of intents for easy indexing and parsing
Re: (Score:2)
It's nothing new.
PocketSphinx is a lightweight speech recognition engine, specifically tuned for handheld and mobile devices, though it works equally well on the desktop
https://github.com/cmusphinx/p... [github.com]
Re:local processing (Score:4, Informative)
And it says right there on the page
"THIS IS A RESEARCH SYSTEM. This is also an early release of a research system. We know the APIs and function names are likely to change, and that several tools need to be made available to make this all complete. "
It also has not been updated in 3 months, so its not extremely active either. Its not surprising its from CMU though, their flite library is widely used for speech synthesis all over the place for text to speech.
Either way, the actual parsing being done on mobile devices is minimal at this point. Apple does not do it, and google does it only in specific circumstances with very basic things like 'play music', 'open gmail', etc. There is no in depth parsing to actually figure out context and meaning without the online component as of yet
Re: (Score:3)
I imagine they will try to make this available offline eventually, because they do with other services that are related to travel. Google Translate (text/OCR via the camera) works offline, and Google Maps lets you download areas for offline viewing, and their new Travel app thingy has offline caching as well.
The new always-on music identification system uses a local database too.
It makes sense to do speech processing on the phone where possible, because it decreases latency. To give the best user experience
Re: (Score:2)
Having tried Sphinx, I feel comfortable saying it is terrible compared to the major players. All of them are doing the speech recognition server side. Nuance is probably the biggest technology vendor, but Google and Microsoft have their own in-house technology.
Re: (Score:2)
Yeah I agree that Sphinx is not that great. My point was merely that it *can* be done on the device.
Re:local processing (Score:4, Interesting)
For the iPhone and iPad Apple still sends the data over to their servers to be processed. If you turn on Siri on the Mac it sends data to Apple as well. However you can turn this off (or at least some of this from going to Apple) by going to the Keyboard preferences and under the Dictation section turning on the Use Enhanced Dictation feature. This allows offline use and does the processing on your computer. It also downloads about 1 GB of data when you turn it on (the first time - I hope it keeps it around if you turn it off).
While I'm sure that the processor is powerful enough to do the work on the phone it would be a big drain on the battery, at least more than the network used. Also storing 1 GB would really get people complaining. And that would probably be per language. I would also imagine the RAM requirements for such a program would be fairly heavy too.
Re: (Score:3)
Just try it.... (Score:4, Informative)
1. Grab an Android phone
2. Open Google translate
3. Tap "Offline translation", pick a language to download
4. Put the phone into airplane mode
5. Tap and talk
Surprise, it works.
Re:Just try it.... (Score:4, Informative)
1. Grab an Android phone
2. Open Google translate
3. Tap "Offline translation", pick a language to download
4. Put the phone into airplane mode
5. Tap and talk
---
Absolutely! As a quick test of my phone shows..
1 Greta Indoor Fun
did it open Google Translate
three attacked offline translator pickup a a language to download
for put the phone in airplane mode
5 tape and talk
Express, it works.
Re: (Score:1)
Grrr.
The "5 tape and talk" is my typo. The phone wrote "5 tap and talk" correctly.
My phone stabs me in the back every... single... day.
It is probably useful for translation- but I wouldn't count on it for anything complicated.
And especially not jargon or idioms.
Re: (Score:1)
Google has been doing "offline" voice recognition directly on phones for a few years now.
http://stackandroid.com/tutori... [stackandroid.com]
Adding a translation layer that also runs locally isn't that far fetched.
Re: (Score:2)
Point is not that what could be run on local device. When all the data is transferred and stored to cloud, it can be monetized now and in the future in new innovative ways.
Good point. And from a global perspective, this volume of data also make huge leaps in AI possible.
Re: (Score:2)
We all know that processing is not done on the phone.
Indeed you're right, unless you simply click the "download" button on the language pack in Google translate allowing you to use all of the features including audio and video live translation offline without any data connection. You know, like a person who is in another country is expected to be able to without incurring roaming charges.
I know sorry, doesn't fit your anti-Google narrative. Damn those pesky facts.
Re: (Score:1)
Roaming charges will go away. The internet is going to be big some day, just be patient.
I'm not sure local processing is going to make sense for machine learning applications. The real-world usage expands the training corpus which means the online services will continue to improve.
on-device or over-internet translation? (Score:1)
Happy ending (Score:5, Funny)
I could see where this could be useful. Just last week, I went to the Thai massage parlor and the girl asked me if I wanted "the works". When I said yes, they gave me a mop and bucket and told me to start in the lobby.
Re: (Score:1)
The march of progress (Score:5, Funny)
This is clearly the most high-tech way yet to say: "My hovercraft is full of eels"
Re: (Score:1)
Re: (Score:3)
Yoshi! (Score:2)
Any steps towards effective universal translators are most welcome!
So normal bluetooth headphones won't work for this (Score:5, Insightful)
This sounds like you need the GOOGLE earbuds to allow translation services, which happens on the phone. Is this just an attempt at vendor lock-in? Or will this work with any bluetooth headphone supporting device? I don't see how it could.
This abandonment of the headphone jack really looks to be rather expensive and inconvenient to the end user. Almost feels like they are copying the inkjet printer refill fiasco.
Re: (Score:1)
Yes you need the Google earbuds. It was done that way for "UX reasons". https://news.ycombinator.com/item?id=15404918 [ycombinator.com]
The earbuds send the data to the phone which sends it to Google which translates everything into "my hovercraft is full of eels" and then sends it back to your phone which sends it to the earbuds.
The only new part here is of course the earbuds.
Re: (Score:2)
Re: (Score:2)
All good reasons on that link... until you remember that none of those reasons would apply if they had kept a 3.5mm headphone socket for wired headphones.
WTF does a 3.5 mm headphone jack have to do with BT earbuds?
If anything, that makes this product MORE compatible across devices and platforms.
Re: (Score:2)
For the translation services, you have to use the Google BT headphones, there are no other compatible options. The reasons they give for making it specific to their headset are all specific to BT headphones. If they had support for a 3.5mm TRRS plug then they would have compatibility with a huge range of other options.
I would think that the real reason is that those earbuds have an App that receives a dedicated command from the earbuds, initiating the translation and text to speech process.
Just guessing, though.
Re: (Score:1, Flamebait)
This sounds like you need the GOOGLE earbuds to allow translation services, which happens on the phone.
Translation does not happen on the phone. All what you hear will be sent to Google, probably primarly for "optimizing advertisements presented to you" and other eavesdropping purposes, and secondarily, Google will attempt some translation sent back to your phone.
Unless, of course, you have no Internet connection or service from Google, in which case there is no translation at all.
Re:So normal bluetooth headphones won't work for t (Score:5, Informative)
Translation does not happen on the phone. All what you hear will be sent to Google
Precisely. Unless you simply download the small language pack in Google translate enabling it to work on video, audio and text completely offline. You know, like when you're in another country without mobile coverage.
Oh what, didn't fit your narrative?
Re: (Score:2)
Re: (Score:2)
Translation does not happen on the phone. All what you hear will be sent to Google
Precisely. Unless you simply download the small language pack in Google translate enabling it to work on video, audio and text completely offline. You know, like when you're in another country without mobile coverage.
Great. Then tell me please the link where to download the the "small language pack" for Greece, where I intended to spend my next holiday. Or is there none to fit your narrative?
Re: (Score:2)
Err the download button is immediately to the right of the language select button, next to the word "Greek".
Silly users don't fit the narrative very well.
Re: (Score:2)
Or will this work with any bluetooth headphone supporting device? I don't see how it could.
I think this may be vendor lock-in. There doesn't seem to be any capability here that isn't part of Google translate with the exception of doing it with a button from the headset.
Re: (Score:2)
No, normal Bluetooth headphones should work, assuming you use your phone a microphone for the other person. The translation service should also work offline as long as you're using an android phone (although, it won't be as good as the online translation).
I think this is just a marketing play because Google noticed that a different manufacturer did something similar with android hardware that one would place around one's neck and charged thousands of dollars for the device.
Repackaged Google Translate Feature (Score:2)
You've been able to do more or less the same demo with an Android Gear watch and phone for a couple of years. The output was just text instead of spoken aloud. Honestly it would be a lot nicer if both people could just speak through a single phone. Their new ability to recognize speakers combined with language detection should be enough for that, you'd think. This just feels awkward. If you could invite a robot translator into a phone call, that would be awesome.
Re: (Score:2)
You didn't need anything fancy, it works through any bluetooth headset.
"Pixel Buds"? (Score:2)
Come on, Google, you know everybody's just going to call it "Babelfish."
Re: (Score:2)
Could be a killer feature, but... (Score:4, Insightful)
Then I remembered that Google Translate mostly churns out total nonsense when going to/from Chinese, and I was less excited.
Re: (Score:1)
If it was just Google Translate's normal gibberish it might just still be barely usable, but when paired with voice recognition the failure will be spectacular. You'll probably have better luck with the tried-and-true 'Keep speaking english, just slower and much louder' technique.
What is actually new here? (Score:2)
I still think its pretty cool, but what is it about these new earbuds that has any bearing on this bidirectional spoken translation capability which has been baked into the Translate app for ages?
Re: What is actually new here? (Score:1)
The difference is you need to buy something new for the same functionality.
Consumerism as slightly more subtle use. 2 people with ear buds instead of holding their phones while talking.
Re: (Score:2)
I bet you still bring your food up to the roof and put it in front of your radar antenna, because that's exactly the same thing as using a microwave...
Languages don't work like that (Score:2, Interesting)
Even for native speakers of multiple languages you have to wait until the end of the sentence before translating. So the TFS is fucking bullshit (again).
Re:Languages don't work like that (Score:5, Insightful)
- They know your location
- They know the weather and all events at the place where you are
- They know your name
- They know where you live
- They may know your friends
- They know what you watched on Youtube
- They know what you have been searching for online
- They know all things mentioned above about the person you are talking to
Combine this with some facial expression recognition, which should be possible with iPhone8, and some more context recognition via camera, and you will have more context than most human translators would have.
Re: (Score:3)
That is just a completely stupid statement. None of that information provides context into a current conversation as they are situational dependent, and not dependent on your personal information.
Translators can get far more context into translating language without knowing any of the above information.
Re: (Score:2)
They get all the context they need from automatically generated data-graphs.
You can't data graph data you don't have.
Re: (Score:2)
So you mean Google does not know the context? Let's see: Ok maybe I didn't make it clear. The order of the words in a sentence differ in different languages, so before you can translate you have to hear the entire sentence in order to accurately translate it. This is what I meant by context. Since you have to wait to hear an entire sentence before translating it can never be real time.
Re: (Score:2)
Anyone who speaks more than one language will tell you that context is one of the biggest hurdles in translating language. It's not just straight word for word translation, the context might not be given until you get to the end of the sentence, so it is actually impossible to translate language in real time.
Even for native speakers of multiple languages you have to wait until the end of the sentence before translating. So the TFS is fucking bullshit (again).
You'd be right if you weren't wrong, or at least partially. Word for word translation often gets you very far through if you have the ability to recognise what is being said through a damaged sentence structure. Google translate has done this for a long time already with its image translation feature.
If you use the live translation feature it will offer word for word replacement. If you take the picture and select the words it attempts to translate with context. Depending on the differences in the languages
Re:Languages don't work like that (Score:4, Insightful)
Yeaaahhh. Google Translate still makes a mess of written Japanese, despite being trained on that.
Conversational Japanese is pretty much impossible to translate well to English on a phrase-by-phrase basis without loads of context. Even for humans. No way an earplug and google translate is going to pull that off.
It would be pretty useful for tourist stuff though. asking prices, directions etc. I've seen some quite good demos of on-the-fly translation in that context.
Re: (Score:2)
It would be pretty useful for tourist stuff though. asking prices, directions etc. I've seen some quite good demos of on-the-fly translation in that context.
This right here is the key bit. Language translation doesn't need to be perfect, or even sensible in order to be incredibly useful.
The only thing it needs to do is make enough sense not to negate meanings. I have on occasion seen that, where a translation has come through with the exact opposite of the original meaning, but that rarely happens through similar languages. But when you look at a country like Europe with 23 official languages, even many more unofficial ones, and yet the vast majority fall into
Re: (Score:2)
Split across the continents though and you start needing context in order to understand the translations. Though it is funny to see Chinese restaurants called "Frank food. Come eat do!"
Funny, but an accurate description of some of the "frank food".
A clear failure (Score:1)
I'm not sure I'd ever trust the translation capability of a device that called themselves "Pixel Earbuds". Those words just don't go together.
The BabelFish has come to pass (Score:4, Insightful)
hmmm (Score:2)
Re: (Score:2)
September 1991 (Score:2)
People will now be hearing all kinds of "Darmok and Jalad at Tanagra", finally realizing how much of what we say is in phrases and not words.
Ahh, their eyes open; their sails unfurl.
The video is unavailable (Score:2)
Anyone got a link to the video that works?
This is a great step (Score:1)
Star Trek has arrived (Score:2)
Re: (Score:2)
Does it do Klingon?
Um, yeah... and the accuracy rate? (Score:3)
First, how does it know, or do you tell it, what language to translate?
And second... I read that red book that led me to want to learn more about lead, and wha's happenin', babeeee?
At work, we have a "hybrid phone, and we get "voicemail previews" "powered by M% speech technology". They range from sorta-kind ok, to Vogon poetry. Now, *I* wouldn't let alpha software out the door, but we are talking M$, where *you* are the (unpaid) beta tester.
Re: (Score:2)
Number of times I wanted to listen to music in the past week? 3.
Number of times I wanted a translation in the past, Oh, I dunno, 50 years? 0.
On the other hand, it'll probably be a lot more useful than any of the new "features" they keep shoehorning into Firefox. (Just added lockPref("browser.newtabpage.activity-stream.enabled", false); to my mozilla.cfg file.)
Re: (Score:2)
The numbers are reversed for me. To each his own.
Re:Lessee... (Score:5, Insightful)
Number of times I wanted a translation in the past, Oh, I dunno, 50 years? 0.
Travel is fatal to prejudice, bigotry, and narrow-mindedness,
Actually, with that in mind, it's probably safest if you continue to stay where you are.
Re: (Score:2)
https://www.youtube.com/watch?v=QKxaxrxVMsg [youtube.com]
Re: (Score:1)
Looking something up on the internet and actual interaction with people with different cultures is not the same thing. Interacting with people makes you think of them on a more personal level, rather than an abstract concept.
Don't get me wrong I hate travel, especially going around at looking at buildings, art etc. But interacting with people from different cultures and getting different opinions and world views I find quite interesting. This may make the interactions easier.
Africans and others are dying o
Re: (Score:3)
Listen to music? About 8 hours a day.
Translations in the last two weeks? A few dozen (China, Malaysia). These would be awesome to have just for the translation capabilities.
Re: (Score:2)
Translations, how about the entirety of the internet that is not English based, every bit of text, every video, not so much music though, as it translate differently https://www.youtube.com/watch?... [youtube.com].
No to get away from the clearly planted diversions, the biggest problem Google, which is a demostated corrupt political player and acting in the most evil ways to pervert democracy in it's investors and directors personal favour, listen in on your conversations and not just listening actively processing those
Re: (Score:2)
Re: (Score:2)
Listen to music? About 8 hours a day.
A babel fish sounds great, but with 5 hours of battery life when brand new, these won't get you through a day with music.
Re: (Score:2, Funny)
All that foreign porn and NOT ONCE have you wondered what they were saying?
Re: (Score:2)
All that foreign porn and NOT ONCE have you wondered what they were saying?
All I know is... French women talk a lot about urinating when they get excited.
Wee... wee....
Re: (Score:2)
Re: (Score:1)
There is a pretty big world outside of your country. You should visit sometime. I use translation nearly every day.
Re: (Score:1)
Let me see, living in Norway, but having a geographically quite spread out company.. in the last year I've travelled to:
- Sweden (2x)
- UK (3-4x)
- Ireland (1x)
- France (1x)
- Italy (1x)
- Spain (10x+)
- Mexico (1x)
- Bulgaria (1x) (vacation)
- Netherlands (1x) (vacataion)
While Sweden, UK and Ireland were entirely fine without translator devices - I absolutely needed to use translation services on a daily basis in the rest of the countries (We
Re: (Score:1)
To be honest -- that's not really all that geographically spread out. Except for Mexico, you could fit all that in a Canadian province almost. Nationally spread out, sure. But all of those places a very close geographically.
Re: (Score:1)
Not really. Canada is just very large.
Re:Lessee... (Score:4, Interesting)
I want this. In San Antonio it's very common for people to switch to Spanish whenever they see a white person come near.
My brother and I made up a nonsensical language that we would switch to in elevators and other places when this would happen. We had no idea what we were saying but at least we helped the other people not be the only jerks in the room :)
Just having it a work would be a dream when my peers and managers switch to Spanish and start giggling and laughing, Funny thing is that they get pissed when people from Korea switch to Korean in front of them
Re: (Score:2)
Ummm... why not just learn Spanish?
Re: (Score:2)
Number of times I wanted to listen to music in the past week? 3.
Number of times I wanted a translation in the past, Oh, I dunno, 50 years? 0.
I dunno, could be useful for listening to the coworkers who switch languages to talk about the juicy things. You know when they stop speaking English and start speaking their native language it must be something juicy.
Re: (Score:2, Insightful)
Re:Lessee... (Score:5, Funny)
My guess is he is a suburbanite who is afraid of meeting new people so never has seen a foreigner.
Fuck you monkey, go back to your own shitty hindustan, I don't want you here.
Guess confirmed. In only 8 minutes!
Re: (Score:3)
There are many settings where this could be very useful outside of the UN. Like any business dealing with a multi-ethnic customer base (and no, I don't mean multi-national corporations only - it could apply to many corner stores in NYC for example). Or that country that you've wanted to visit but you don't speak the language of, and they don't speak English.
Re: (Score:3)
Re: (Score:2)
From what I've seen in stand-up comedy and TV shows, it could be useful for Americans wanting to take a vacation in the UK.
Why is it that Americans moan about hard to understand UK dialects/accents, whereas no one here has any problem watching US TV shows whether they're set in Alaska, New York or Texas? Is American English just more homogeneous?
Re: (Score:2)
Re: (Score:3)
Don't be numpty and spread such porkies la.
Re: (Score:2)
Don't be so fricken sensitive. It was clearly a joke.
It's a pretty established one too. Simpsons and Arrested Development come to mind with several jokes on this.
Honestly, we have trouble understanding anything we are not exposed to consistently. I watch allot of Dr Who other BBC shows and have no trouble with most the accents there but get stuck hearing a US northeastern accent.
Re: (Score:3)
From what I've seen in stand-up comedy and TV shows, it could be useful for Americans wanting to take a vacation in the UK.
Why is it that Americans moan about hard to understand UK dialects/accents, whereas no one here has any problem watching US TV shows whether they're set in Alaska, New York or Texas? Is American English just more homogeneous?
Actually... yes.
America has it's own regional dialects but it's nothing like Britain; with some exceptions, you can travel two or three states and not notice a change in accent. It's not like Britain where if you drive 30 miles from Merseyside to Manchester and the accent is vastly different.
There's certainly a difference between a Boston accent and an Alabaman accent but it's not nearly as different as Geordie to west country.
Re: (Score:2)
Re: (Score:2)
Don't tell me you've never heard a True Scotsman speak? It's practically unintelligible, until you get used to it.
Re:Babelfish (Score:5, Informative)
Re: (Score:1)
Your confusion stems from the fact that the ambiguity of the word yellow. When Douglas Adams wrote that bit, he was using a particular dialect of English in which the words "yellow" and "orange" are basically the same, and only the context gives you any clue as to precisely which color is being referred to.
Without any real context, it's impossible to tell whether he meant yellow as in the yellow of a fresh lemon, or the yellow of the sun, (which only appears as yellow as it does because of its sheer intens
Re: (Score:2)
Re: (Score:2)
Re: (Score:3)
Can't believe it took this long for a babelfish reference - and then you go and muck it up.
Re: (Score:2)