Goto

Collaborating Authors

 Asia


End-to-End Waveform Utterance Enhancement for Direct Evaluation Metrics Optimization by Fully Convolutional Neural Networks

arXiv.org Machine Learning

Speech enhancement model is used to map a noisy speech to a clean speech. In the training stage, an objective function is often adopted to optimize the model parameters. However, in most studies, there is an inconsistency between the model optimization criterion and the evaluation criterion on the enhanced speech. For example, in measuring speech intelligibility, most of the evaluation metric is based on a short-time objective intelligibility (STOI) measure, while the frame based minimum mean square error (MMSE) between estimated and clean speech is widely used in optimizing the model. Due to the inconsistency, there is no guarantee that the trained model can provide optimal performance in applications. In this study, we propose an end-to-end utterance-based speech enhancement framework using fully convolutional neural networks (FCN) to reduce the gap between the model optimization and evaluation criterion. Because of the utterance-based optimization, temporal correlation information of long speech segments, or even at the entire utterance level, can be considered when perception-based objective functions are used for the direct optimization. As an example, we implement the proposed FCN enhancement framework to optimize the STOI measure. Experimental results show that the STOI of test speech is better than conventional MMSE-optimized speech due to the consistency between the training and evaluation target. Moreover, by integrating the STOI in model optimization, the intelligibility of human subjects and automatic speech recognition (ASR) system on the enhanced speech is also substantially improved compared to those generated by the MMSE criterion.


AI wave rolls through Microsoft's language translation technologies

#artificialintelligence

A fresh wave of artificial intelligence rolling through Microsoft's language translation technologies is bringing more accurate speech recognition to more of the world's languages and higher quality machine-powered translations to all 60 languages supported by Microsoft's translation technologies. The advances were announced at Microsoft Tech Summit Sydney in Australia on November 16. "We've got a complex machine, and we're innovating on all fronts," said Olivier Fontana, the director of product strategy for Microsoft Translator, a platform for text and speech translation services. As the wave spreads, he added, these machine translation tools are allowing more people to grow businesses, build relationships and experience different cultures. Microsoft's research labs around the world are also building on top of these technologies to help people learn how to speak new languages, including a language learning application for non-native speakers of Chinese that also was announced at this week's tech summit. The new Microsoft Translator advances build on last year's switch to deep neural network-powered machine translations, which offer more fluent, human-sounding translations than the predecessor technology known as statistical machine translation.


China to lead in healthcare breakthroughs - Chinadaily.com.cn

#artificialintelligence

China leads the pack in the next healthcare breakthrough with the government and stalwart innovators bankrolling a big effort in artificial intelligence and big data that could essentially blaze a new trail in the sector, says an expert. "For years, the United States has been seen as the impeccable frontrunner for a big bang coming in the healthcare sector, in terms of money invested in digital healthcare, as well as the number of groundbreaking startups," said Sigal Atzmon, president of Medix Group, an international company that offers medical case management. "When it comes to the implementation of trailblazing technologies in the healthcare sector, however, China is the clear leader," Atzmon noted. Unlike the United States, where people have the "whole existing infrastructure legacy" that must be taken down and an awfully lot of stakeholders involved who may not favor a change, China has much more freedom to chart its own course and implement the game-changing technologies thanks to the "late-mover advantage", or how innovative newcomers outperform pioneers, she said. The Belgium-born senior corporate executive, who founded Medix in 2006 after her own scary experience with a breast mammogram, believes AI technology and digital healthcare stand as the unique and cost-effective answer for the country's under-doctored, underfunded and overburdened hospitals, where patients reportedly have to wait a long time or even pay large sums of money to middlemen to get appointments with renowned specialists.


Alicia Vikander makes a persuasive Lara Croft in the new-and-somewhat-improved 'Tomb Raider'

Los Angeles Times

A message concealed in one of Richard's old puzzle-boxes grants Lara access to her father's archive of materials on the Himiko project. Tasked with finding out how he met his end, she jets off to Hong Kong, where she befriends a drunk, rifle-toting boatman, Lu Ren (Daniel Wu), who eventually sobers up and agrees to steer her through tempestuous waters to Yamatai. This may be as good a place as any to note that "Tomb Raider" was directed by Norway's Roar Uthaug, who previously directed the 2016 disaster flick "The Wave," and who once again proves a peerless choreographer of massive walls of computer-generated water.


Robotic Tortoise Helps Kids to Learn That Robot Abuse Is a Bad Thing

IEEE Spectrum Robotics

Kids like to touch things. Kids like to whack things. This is usually fine when the thing is a toy, but it can be a problem when the thing is a robot. We've written about children beating robots up before, and it seems like it's an inevitability when kids (or even some adults) meet a robot for the first time: They want to see what it can do and how it reacts to things, and that can result in some behaviors and interactions that would be pretty upsetting if they were targeted at something alive. That is to say, sometimes kids are abusive towards robots, especially when there aren't any consequences to the things that they do.


How AI Is Addressing the Fraud in Advertising: Here's What You Need to Know

#artificialintelligence

Given the online advertising boom in today's digital age, businesses need to learn to allocate their marketing dollars wisely. Unfortunately, there's a big challenge to meeting that goal: Advertising fraud and lackluster results are rife in the digital ad space. As a business owner, digital manager and entrepreneur, you've likely already experienced these issues, via paid Instagram ads that failed to achieve their intended results, or paid services like Google Adwords that had no conversions. To avoid such outcomes, you need to make sure your ads stand out and that you're spending your dollars on actual results -- not bot traffic. To clarify my Instagram example: Let's say that you paid Instagram to promote your post, and, all of a sudden, you start gaining a bunch of followers who, according to their account information, live in India and have only about five pictures each on their individual profiles. I'm no ad-fraud detective, but judging from the looks of that kind of new traffic, you'd be wise to question these followers' validity.


Dubai Decrees Itself the A.I. City State of the Future

#artificialintelligence

While the US government seem not yet convinced that AI power would rule the world, China and Dubai appear dead serious about the technology. In fact, Dubai a key member of the Emirates has revealed its plans to grab "soft power," in print, under the new ministry of AI. Yes, you hard that right, artificial intelligence is now a complete cabinet docket that receives government funding like health, security, and other serious ministries. It's this office that keeps the print, which reveals how autonomous robocops might in future patrol around the Dubai Mall. How smart drones will deliver goods to addresses, how flying taxis will lift and drop commuters around the city, and how buslike pods with brains will pick and drop passengers from their doorsteps.


Almost a third of consumers plan for new AI home devices

#artificialintelligence

Almost a third (32%) of consumers surveyed globally by PwC plan to buy an AI device including robots or automated assistants, with retailers watching closely as'voice commerce' develops in the home. The findings are published today in PwC's Global Consumer Insights survey, which assesses the shopping behaviour, habits and expectations of over 22,000 consumers in 27 countries. The study reports that 10% of respondents already own artificial intelligence (AI) devices, such as robots and automated personal assistants like Amazon Echo or Google Home, and 32% said they plan to buy one. Both consumer and retailer habits and offerings still need time to adapt however, to make the most of the new voice commerce channel. Interest in the devices is strongest amongst consumers in emerging economies including China, Vietnam, Indonesia and Thailand.


AI drones are controlling self-driving diggers on building sites

New Scientist

Drones soar in the skies while scanning the ground. In the dirt below, huge diggers are working semi-autonomously, levelling land and digging ditches. Californian firm Skycatch has supplied its quadcopter drones to more than 5000 building sites in Japan over the past three years. The sites are mostly in and around the Tokyo area and are run by Komatsu, the world's second-largest building firm, as part of its Smart Construction project.


Google brands cryptocurrency adverts 'deceptive' and bans them from search results

The Independent - Tech

Google has announced it will ban all advertising relating to cryptocurrency products and services from its search engine from June in a bid to tackle "emerging threats". The company said in a blog post it was cracking down on "deceptive content" including commercials for all and any speculative and unregulated financial products like cryptocurreny exchanges, wallets and initial coin offerings as well as binary options, foreign exchange trading platforms and spread betting sites in a bid to improve the "ad ecosystem" experienced by users. Facebook came to the same decision in January, snubbing crypto ads across the social media site plus Instagram and Audience Network. Cryptocurrencies have continued to make headlines this year after an explosion in the price of bitcoin, ether and their kind in the run-up to Christmas, with bitcoin's value spiking at $19,850 (£14,214) in mid-December. The market is currently valued at $352bn (£252bn), according to CoinMarketCap, with over 1,500 types of virtual coin in play.