Mozilla releases dataset and model to lower voice-recognition barriers

By -
The browser maker has collected nearly 500 hours of speech to help voice-recognition projects get off the ground.


Mozilla has released its Common Voice collection, which contains almost 400,000 recordings from 20,000 people, and is claimed to be the second-largest voice dataset publicly available.
The voice samples in the collection were obtained from Mozilla's Common Voice project, which allowed users via an iOS app or website to donate their utterances. It is hoped that creating a large public dataset will allow for better voice-enabled applications.
"One reason so few services are commercially available is a lack of data," Mozilla senior vice president of emerging technologies Sean White said in a blog post.
"Startups, researchers, or anyone else who wants to build voice-enabled technologies need high-quality, transcribed voice data on which to train machine-learning algorithms. Right now, they can only access fairly limited data sets."
At the moment, the collection is focused on English, but there are plans to extend it to other languages in the first half of 2018.
Alongside its dataset, Mozilla also released its open-source Project DeepSpeech voice-recognition model based on work done by Chinese internet giant Baidu. It is claimed that with its 6.5 percent error rate on the LibriSpeech dataset, DeepSpeech is approaching human levels of recognition.
In August, Microsoft said it had reached a voice-recognition error rate of 5.1 percent on the Switchboard corpus, the same level as professional human transcribers.
Despite the new milestone, Microsoft acknowledges that machines still find it tough to recognise different accents and speaking styles, and don't perform well in noisy conditions.
Earlier in the year, Google said it had a 4.9 percent error rate in its speech-recognition software.
Samsung has said it is looking to use voice recognition throughout its home appliance line-up by 2020, and recently partnered with Kakao to cooperate on AI and voice recognition.

Comments

Post a Comment

Popular Posts

Hacker steals data of millions of Bulgarians, emails it to local media

By -
Image
  Source of the data breach appears to be the country's National Revenue Agency A mysterious hacker has stolen the personal details of millions of Bulgarians and has emailed download links to the stolen data to local news publications. The data's origin is believed to be the country's National Revenue Agency (NRA), a department of the Bulgarian Ministry of Finance. In a  message posted on its website  on Monday, the NRA said it was working with the Ministry of the Interior and the State Agency for National Security (SANS) to investigate the hack. "We are currently verifying whether the data is real," said the NRA. Hours after this article's publication, the Bulgarian Ministry of the Interior  confirmed the hack . HACKER STOLE 110 DATABASES, LEAKED 57 According to reports from local media [ 1 ,  2 ,  3 ,  4 ,  5 ], who received part of the data, the hacker said they stole the personal details of over five million Bulgarians, of the country's total populatio
Read more

​Linux totally dominates supercomputers

By -
Image
It finally happened. Today, all 500 of the world's top 500 supercomputers are running Linux. Linux rules supercomputing. This day has been coming since 1998, when Linux first appeared on the TOP500 Supercomputer list . Today it finally happened:  All 500 of the world's fastest supercomputers are running Linux . The last two non-Linux systems, a pair of Chinese IBM POWER computers running AIX, dropped off the  November 2017 TOP500 Supercomputer list . Overall, China now leads the supercomputing race with 202 computers to the US' 144. China also leads the US in aggregate performance. China's supercomputers represent 35.4 percent of the Top500's flops, while the US trails with 29.6 percent. With an anti-science regime in charge of the government, America will only continue to see its technological lead decline. When the  first Top500 supercomputer list was compiled in June 1993 , Linux was barely more than a toy. It hadn't even adopted Tux as its masc
Read more

Microsoft tries to stem its self-made collaboration-tool confusion

By -
Image
Microsoft is using this week's Ignite conference to try to help clarify its collaboration-tool strategy. Here's how SharePoint, Teams and Yammer figure in the mix. Choice is good. But too much choice, especially when it comes to collaboration tools, has been a problem for Microsoft. This isn't news to customers, partners or Microsoft execs themselves. But at the company's Ignite IT Pro conference in Orlando this week, Microsoft execs took a step to try to clarify the company's strategy and messaging in this area. Microsoft Office 365 Marketing chief Ron Markezich kicked off the conference this week with a slide entitled "Microsoft 365 Teamwork: Where to Start a Conversation." That slide attempts to do what  Microsoft initially attempted with a 60-plus-page whitepaper : Clarify which collaboration tools customers should use when. The slide, which features SharePoint -- and its files, sites and content storage at the center -- is broken down into t
Read more