adplus-dvertising
Connect with us

Tech

Why Google Assistant supports so many more languages than Siri, Alexa, Bixby, and Cortana – VentureBeat

Published

 on


Google Assistant, Apple’s Siri, Amazon’s Alexa, and Microsoft’s Cortana recognize only a narrow slice of the world’s most widely spoken languages. It wasn’t until fall 2018 that Samsung’s Bixby gained support for German, French, Italian, and Spanish — languages spoken by over 600 million people worldwide. And it took years for Cortana to become fluent in Spanish, French, and Portuguese.

But Google — which was already ahead of the competition a year ago with respect to the number of languages its assistant supported — pulled far ahead this year. With the addition of more than 20 new languages in January 2019, and more recently several Indic languages, Google Assistant cemented its lead with over 40 languages in well over 80 countries, up from eight languages and 14 countries in 2017. (Despite repeated requests, Google would not provide an exact number of languages for Google Assistant.) That’s compared with Siri’s 21 supported languages, Alexa’s and Bixby’s seven languages, and Cortana’s eight languages.

So why has Google Assistant pulled so far ahead? Naturally, some of the techniques underpinning Google’s natural language processing (NLP) remain closely guarded trade secrets. But the Mountain View company’s publicly available research sheds some — albeit not much — light on why rivals like Amazon and Apple have yet to match its linguistic prowess.

300x250x1

Supporting a new language is hard

Adding language support to a voice assistant is a multi-pronged process that requires considerable research into speech recognition and voice synthesis.

Most modern speech recognition systems incorporate deep neural networks that predict the phonemes, or perceptually distinct units of sound (for example, p, b, and d in the English words pad, pat, and bad). Unlike older techniques, which relied on hand-tuned statistical models that calculated probabilities for combinations of words to occur in a phrase, neural nets derive characters from representations of audio frequencies called mel-scale spectrograms. This reduces error rates while partially eliminating the need for human supervision.

Speech recognition has advanced significantly, particularly in the past year or so. In a paper, Google researchers detailed techniques that employ spelling correction to reduce errors by 29%, and in another study they applied AI to sound wave visuals to achieve state-of-the-art recognition performance without the use of a language model.

Parallel efforts include SpecAugment, which achieves impressively low word error rates by applying visual analysis data augmentation to mel-scale spectrograms. In production, devices like the Pixel 4 and Pixel 4 XL (in the U.S., U.K., Canada, Ireland, Singapore, and Australia) feature an improved Google Assistant English language model that works offline and processes speech at “nearly zero” latency, delivering answers up to 10 times faster than on previous-generation devices.

Of course, baseline language understanding isn’t enough. Without localization, voice assistants can’t pick up on cultural idiosyncrasies, or worse they run the risk of misappropriation. It takes an estimated 30 to 90 days to build a query-understanding module for a new language, depending on how many intents it needs to cover. And even market-leading smart speakers from the likes of Google and Amazon have trouble understanding certain accents.

Google’s increasingly creative approaches promise to close the gap, however. In September, scientists at the company proposed a speech parser that learns to transcribe multiple languages while at the same time demonstrating “dramatic” improvements in quality, and in October they detailed a “universal” machine translation system trained on over 25 billion samples that’s capable of handling 103 languages.

This work no doubt informed Google Assistant’s multilingual mode, which, like Alexa’s multilingual mode, recognizes up to two languages simultaneously.

Speech synthesis

Generating speech is just as challenging as comprehension, if not more so.

While cutting-edge text to speech (TTS) systems like Google’s Tacotron 2 (which builds voice synthesis models based on spectrograms) and WaveNet 2 (which builds models based on waveforms) learn languages more or less from speech alone, conventional systems tap a database of phones — distinct speech sounds or gestures — strung together to verbalize words. Concatenation, as it’s called, requires capturing the complementary diphones (units of speech comprising two connected halves of phones) and triphones (phones with half of a preceding phone at the beginning and a succeeding phone at the end) in lengthy recording sessions. The number of speech units can easily exceed a thousand.

Another technique — parametric TTS — taps mathematical models to recreate sounds that are then assembled into words and sentences. The data required to generate those sounds is stored in the parameters (variables), and the speech itself is created using a vocoder, which is a voice codec (a coder-decoder) that analyzes and synthesizes the output signals.

Still, TTS is an easier problem to tackle than language comprehension — particularly with deep neural networks like WaveNet 2 at speech engineers’ disposal. Translatotron, which was demoed last May, can translate a person’s voice into another language while retaining their tone and tenor. And in August, Google AI researchers showed that they could drastically improve the quality of speech synthesis and generation using audio data sets from both native and non-native English speakers who have neurodegenerative diseases and techniques from Parrotron, an AI tool for people with impediments.

In a related development, in a pair of papers Google researchers recently revealed ways to make machine-generated speech sound more natural. In a study coauthored by Tacotron co-creator Yuxuan Wang, transfer of things like stress level were achieved by embedding style from a recorded clip of human speech. As for the method described in the second paper, it identified vocal patterns to imitate speech styles like those resulting from anger and tiredness.

How language support might improve in the future

Clearly, Google Assistant has progressed furthest on the assistant language front. So what might it take to get others on the same footing?

Improving assistants’ language support will likely require innovations in speech recognition, as well as NLP. With a “true” neural network stack — one that doesn’t rely heavily on language libraries, keywords, or dictionaries — the emphasis shifts from grammar structures to word embeddings and the relational patterns within word embeddings. Then it becomes possible to train a voice recognition system on virtually any language.

Amazon appears to be progressing toward this with Alexa. Researchers at the company managed to cut down on recognition flubs by 20% to 22% using methods that combined human and machine data labeling, and by a further 15% using a novel noise-isolating AI and machine learning technique. Separately, they proposed an approach involving “teaching” language models new tongues by adapting those trained on one language to others, in the process reducing the data requirement for new languages by up to 50%.

Separately, on the TTS side of the equation, Amazon recently rolled out neural TTS tech in Alexa that improves speech quality by increasing naturalness and expressiveness. Not to be outdone, the latest version of Apple’s iOS mobile operating system, iOS 13, introduces a WaveNet-like TTS technology that makes synthesized voices sound more natural. And last December Microsoft demoed a system — FastSpeech — that speeds up realistic voice generation by eliminating errors like word skipping.

Separately, Microsoft recently open-sourced a version of Google’s popular BERT model that enables developers to deploy BERT at scale. This arrived after researchers at the Seattle company created an AI model — a Multi-Task Deep Neural Network (MT-DNN) — that incorporates BERT to achieve state-of-the-art results, and after a team of applied scientists at Microsoft proposed a baseline-besting architecture for language generation tasks.

Undoubtedly, Google, Apple, Microsoft, Amazon, Samsung, and others are already using techniques beyond those described above to bring new languages to their respective voice assistants. But some had a head start, and others have to contend with legacy systems. That’s why it will likely take time before they’re all speaking the same languages.

Let’s block ads! (Why?)

728x90x4

Source link

Continue Reading

Tech

Ask Andy: How can you tell whether a startup is a good place to work? When is it safe to disclose a mental-health challenge to coworkers? – Yahoo Canada Finance

Published

 on


Welcome to the inaugural edition of Ask Andy. In this biweekly column, Andy Dunn—the founding CEO of Bonobos and Pie—offers advice on leading teams, building things, and surviving the startup life. Got a question for Andy? Ask it here.

***

As a software developer who would like to work for a startup, what should I look for in a company so that I know it’s legit? If I am putting a lot of work into a product, I want to know that at minimum it’s for a legitimate company and founder—not just another person with an overdone app idea that knows nothing about the tech world. Sarah C.

300x250x1

If you’re learning the startup game, the best bet here is to go later-stage. Focus on a pre-IPO company that is growing quickly, has raised money from blue-chip investors, and is getting positive buzz in the market that it will go public within the next two years.

ADVERTISEMENT

Then, don’t believe any of it.

Network your way into three of the company’s team members on LinkedIn or through your network. Have three virtual or IRL coffees. Have them tell you about the culture: If they’re learning; if the company’s really growing; and most importantly, whether or not they respect and, ideally, admire the leadership.

Keep looking until you find this vetted opportunity.

That’s a systematic, rational approach. But that’s not the only way to go. You could throw it all out the window.

Find a company where you believe in the mission. One where you fall in love with the product or service. You might already be a high LTV customer or a power user. Check your credit card statement and your app home screen to source ideas. Your passion for the mission will make it work for you for some time, even if the company doesn’t work in the long run.

However you get there, once you’re inside for a year or two, you’ll be learning.

You may have to switch horses. That’s okay.

When you do, you’ll know more people, you’ll have more insight, and the path on what to pick next will be clearer. Heck, you might even notice an inflection point and meet a cofounder that leads to you starting a company yourself.

It’s like dating.

You probably won’t marry your first love—but you might. If you don’t, your judgment will iteratively improve. And the good news is unlike a marriage, you can change out your partner every few years. (What I’ve found, though, is that the most successful people professionally, and those who generate the most wealth, have more like 5- to 10-year runs.)

Trust your intuition. Follow your heart on the mission or product. Then, don’t trust yourself.  Study the market. Use the product.  And do at least three off-list references outside of who you interview with. Read every single Glassdoor entry.

And then jump!

You’ll be fine.

Do you think you could have shared your mental health conditions publicly BEFORE you were professionally successful, and still have been successful? Or was the fact that you had already achieved professional success what allowed you to be open? Zack

No, I don’t think I could have shared before we succeeded. I wouldn’t have had the courage to, and I feared it might be career-limiting.

Then again, it was almost seven years ago that I had my I-can’t-deny-this-any-longer moment with my Bonobos colleagues and investors. As of today, I think it’s becoming more possible to be candid about mental health. I hope we can move to a world where I could have been more open, sooner, at least selectively with my leadership team and board.

Some entrepreneurs ask me when to tell their VCs about the mental-health challenge or mental-health diagnosis they wrestle with. I always say the same thing: at a breakfast meeting, four months after you’ve closed the round and hit your numbers. Nobody cares about your neurodivergence if you’re performing—and most VCs actually know enough to know that most founders have more going on than meets the eye.

With your team, I think it’s doable, even now. Perhaps especially now. The truth is, they know. They know you deal with stuff because they’re around you. And the vulnerability you share in disclosing will multiply their respect for you. More importantly, it’ll give those team members the space to reciprocally share their stuff with their colleagues, and potentially you as well, and bring their full selves to work.

Wouldn’t that be cool?

Andy Dunn is the founding CEO of Bonobos and Pie and the author of Burn Rate: Launching a Startup and Losing My Mind.

This story was originally featured on Fortune.com

Adblock test (Why?)

728x90x4

Source link

Continue Reading

Tech

Take-Two Buys Gearbox And Its New ‘Borderlands’ Game From Embracer – Forbes

Published

 on


If you’re a game developer owned by Embracer Group at this point, you are nervous about layoffs, shutdowns or game cancellations after the last few years. But now, there is a somewhat happy ending for one of them, Gearbox.

It’s just been announced that Take-Two, which owns GTA developer Rockstar, will purchase Gearbox for $460 million. This also includes the properties Gearbox owns, the Borderlands and Tiny Tina’s Wonderlands franchises, Homeworld, Risk of Rain, Brothers in Arms and Duke Nukem. The report says Gearbox has six games in development, five sequels, including a new Borderlands game, the not-announced-but-definitely-happening Borderlands 4. Here’s Strauss Zelnick:

“Our acquisition of Gearbox is an exciting moment for Take-Two and will strengthen our industry-leading creative talent and portfolio of owned intellectual property, including the iconic Borderlands franchise,” said Zelnick, Chairman and CEO of Take-Two. “This combination enhances the financial profile of our existing projects with Gearbox and unlocks the opportunity for us to drive increased long-term growth by leveraging the full resources of Take-Two across all of Gearbox’s exciting initiatives.”

300x250x1

Gearbox has been working with 2K and Take-Two for decades, so it was a logical place for them to land. This is, of course, not a great look for Embracer, who only purchased Gearbox three years ago. The price tag back then was “worth up to $1.3 billion” but there were a lot of strings attached to that where it’s not necessarily the case that selling for $$460 million netted them a ~$900 million loss.

As for what this means for gamers, it would seem something like the Borderlands franchise is now on more stable ground, as it was hard to believe any project at Embracer is fully safe these days. Last year, Embracer quietly cancelled 29 different unannounced games and shut down seven studios in a six month period including Volition and Free Radical Design. That came with around 1,400 layoffs. More recently, Embracer laid off 97 people at Eidos in Janaury and cancelled a Deus Ex game.

Sufficed to say, those at Gearbox probably feel pretty good about this. And as for Take-Two, Borderlands is still a valuable IP, and Tiny Tina’s Wonderland was a surprise hit. There’s a new Homeworld game coming as well. In an era for multi-billion dollar acquisition, Gearbox for $460 million doesn’t seem that bad. That’s probably a third of what GTA 6 will sell on day one next year.

!function(n) if(!window.cnxps) window.cnxps=,window.cnxps.cmd=[]; var t=n.createElement(‘iframe’); t.display=’none’,t.onload=function() var n=t.contentWindow.document,c=n.createElement(‘script’); c.src=’//cd.connatix.com/connatix.playspace.js’,c.setAttribute(‘defer’,’1′),c.setAttribute(‘type’,’text/javascript’),n.body.appendChild(c) ,n.head.appendChild(t) (document);
(function() function createUniqueId() return ‘xxxxxxxx-xxxx-4xxx-yxxx-xxxxxxxxxxxx’.replace(/[xy]/g, function(c) 0x8); return v.toString(16); ); const randId = createUniqueId(); document.getElementsByClassName(‘fbs-cnx’)[0].setAttribute(‘id’, randId); document.getElementById(randId).removeAttribute(‘class’); (new Image()).src = ‘https://capi.connatix.com/tr/si?token=d1021730-df4b-4127-8be2-fb6a0e4e96e4’; cnxps.cmd.push(function () cnxps( playerId: ‘d1021730-df4b-4127-8be2-fb6a0e4e96e4’).render(randId); ); )();

Follow me on Twitter, Threads, YouTube, and Instagram.

Pick up my sci-fi novels the Herokiller series and The Earthborn Trilogy.

Adblock test (Why?)

728x90x4

Source link

Continue Reading

Tech

What's Brewing in the iPhone 16 Rumor Mill? AI, Action Buttons and More – CNET

Published

 on


As the iPhone 15 settles into the market, the tech community is buzzing with anticipation for Apple’s next-generation handset, which is expected to be named “iPhone 16.” 

We’ve heard whispers about the iPhone 16’s features, which are said to span from a new power-efficient display to larger screens, better zoom lenses, an action button and, perhaps not surprisingly, a suite of new gen-AI powered features.

Read more: Best iPhone of 2024

300x250x1

However, the iPhone 16 is still presumably six months away and nothing will be confirmed until Apple’s iPhone event in the fall. Still, these rumors could give us an idea of what to expect from the next iPhone.

Here are the most credible rumors for the iPhone 16.

Will the iPhone 16 fold?

Probably not. The newest rumors suggest Apple has been working on iPhone Flip models in two different sizes, though there have been difficulties in making the devices to Apple’s standards. The company may also be working on a folding tablet with a screen around the size of an iPad Mini. Even though virtually every major phone-maker — from Google to Oppo to OnePlus and Samsung — have launched their own bendable handsets, Apple has been characteristically quiet about whether there will ever be an iPhone Flip or an iPhone Fold.

Prior rumors said Apple may not launch its own flexible screen device until 2025. Samsung hasn’t let phone fans forget it — by releasing an app that will let Apple phone owners experience a Z Fold-esque experience by placing two iPhones side-by-side.

iPhone 16 Pro models to get bigger screens?

Apple has maintained the two screen sizes for iPhone Pro models since 2020 when it launched the 6.1-inch iPhone 12 Pro and the 6.7-inch iPhone 12 Pro Max. However, that’s rumored to change with the iPhone 16 Pro models, which might get bigger screens.

Display analyst Ross Young suggested earlier this year that the iPhone 16 Pro models will have larger screens, putting the sizes at 6.3 inches for the iPhone 16 Pro and 6.9 inches for the iPhone 16 Pro Max. That rumor was later corroborated by Bloomberg’s Mark Gurman, who said the iPhone 16 Pro models could grow by “a couple tenths of an inch diagonally.”

The iPhone 16 and iPhone 16 Plus models are believed to be sticking with the current 6.1-inch and 6.7-inch sizes. If the size increase is accurate, it would be yet another move from Apple to distinguish its Pro iPhone models from its regular ones.

iPhone 15 screen sizes

  • iPhone 15: 6.1 inches.
  • iPhone 15 Plus: 6.7 inches.
  • iPhone 15 Pro: 6.1 inches.
  • iPhone 15 Pro Max: 6.7 inches.

Rumored iPhone 16 screen sizes

  • iPhone 16: 6.1 inches.
  • iPhone 16 Plus: 6.7 inches.
  • iPhone 16 Pro: 6.3 inches.
  • iPhone 16 Pro Max: 6.9 inches.

iPhone 16 gets more AI tricks

One of the most salient selling points of Samsung’s Galaxy S24 series and Google’s Pixel 8 lineup were each of their souped-up AI tips and tricks, and it wouldn’t be a major shock if Apple went in the same direction. Apple CEO Tim Cook has gone on the record this year confirming Apple sees “a huge opportunity for Apple with gen AI and AI.”

According to Gurman’s Power On newsletter, iOS 18 will feature generative AI technology that “should improve how both Siri and the Messages app can field questions and auto-complete sentences.”

A September report from the Information says Apple plans to use large language models, a crucial part of generative AI, to make Siri smarter. The report said this feature is expected to be released with an iPhone software update next year. 

Read More: iPhone iOS 18: A Possible Big Leap In AI

iPhone 16 design: New action button?

In March, AppleInsider published a collection of photographs purportedly displaying 3D-printed dummy models of the rumored iPhone 16 and iPhone 16 Pro. The images revealed that the iPhone 16 may have a vertical camera stack as opposed to a diagonal one and an action button, similar the one on last year’s iPhone 15 Pro.

iPhone 16 gets more power-efficient display?

Another change that could make its way to iPhone 16 displays is greater power efficiency. Samsung Display is apparently developing a new material set, dubbed M14, specifically for Apple, according to a TheElec report, which says the new technology should arrive on iPhones launching next year. M14 will replace the blue fluorescent technology that’s used now with blue phosphorescence technology, creating an even more power-efficient screen than the current LTPO ones used on Pro models, the report says.

Andrew Lanxon/CNET

iPhone 16 gets better zoom?

Both the iPhone 16 Pro and the iPhone 16 Pro Max could both have 5x telephoto lenses next year. According to Apple analyst Ming Chi Kuo, a tetraprism lens will make its way to both Pro models next year, as opposed to just the Pro Max model. Apple equipped the iPhone 15 Pro with a 12-megapixel 3x optical zoom, while the iPhone 15 Pro Max has a 12-megapixel 5x optical zoom camera, which is the equivalent of 120mm lens on a full-frame camera.

If this rumor is true, it could mark a breakthrough in design. When Apple launched the 15 Pro Max with its 5x telephoto lens, it cited the phone’s bigger body as to why the Pro Max had it, but the Pro didn’t.

iPhone 16 processors: A18 chip for all models?

In a break with the past two years, all four iPhone 16 models will apparently get a next-generation Apple chipset, which will all receive A18 branding. According to a MacRumors report citing Jeff Pu, an executive analyst for Haitong International Securities, all four models will have an A18 series chip with Pro iPhone models getting an A18 Bionic Pro and base models getting a regular A18.

The iPhone 15 and 15 Plus currently have an A16 Bionic, which debuted on the 2022 iPhone 14 Pro and 14 Pro Max, while the 15 Pro and 15 Pro Max run on the A17 Pro processor. Pu says the A18 chip will be manufactured with TSMC’s cutting-edge 3 nanometer process.

Read more: Apple iPhone SE 4 Rumors: iPhone 14 Design, Face ID and More



728x90x4

Source link

Continue Reading

Trending