Skip to main content

Microsoft Says Its Speech Recognition System Achieves New Accuracy Milestone

Microsoft's conversational speech recognition system - designed to accurately recognises the words in a conversation like humans do - has reached a 5.1 percent error rate, its lowest so far.

This milestone means that, for the first time, a computer can recognise the words in a conversation as well as a person would.

"Our research team reached that 5.1 percent error rate with our speech recognition system, a new industry milestone, substantially surpassing the accuracy we achieved last year," Microsoft said in a blog post late on Sunday.

Last year in October, the team from Microsoft Artificial Intelligence and Research reported a speech recognition system that makes the same or fewer errors than professional transcriptionists.

The researchers had then reported a word error rate (WER) of 5.9 percent.

"Last year, Microsoft's speech and dialog research group announced a milestone in reaching human parity on the 'Switchboard' conversational speech recognition task, meaning we had created technology that recognised words in a conversation as well as professional human transcribers," said Xuedong Huang, Technical Fellow, Microsoft.

'Switchboard' is a corpus of recorded telephone conversations that the speech research community has used for more than 20 years to benchmark speech recognition systems.

The task involves transcribing conversations between strangers discussing topics such as sports and politics.

The team used "Microsoft Cognitive Toolkit 2.1" (CNTK), the most scalable deep learning software available, for exploring model architectures.

Additionally, Microsoft's investment in cloud compute infrastructure, specifically Azure GPUs, helped improve the effectiveness and speed.

Reaching human parity with an accuracy on par with humans has been a research goal for the last 25 years.

"Microsoft's willingness to invest in long-term research is now paying dividends for our customers in products and services such as Cortana, Presentation Translator, and Microsoft Cognitive Services," the post read.

"Moving from recognizing to understanding speech is the next major frontier for speech technology," the post added.
#CSism #CSismTechnologies #TechnologyTonic #Tech #TechnologyNews
Labels:
Location: B/102, 1st Floor, Ambo Patil Estate, Old VIVA College Road, Above Canara Bank, Opp. Reliance Mall, Virar, Maharashtra 401303, India

Comments

Post a Comment

Thank you for Reading our BLOG

Popular posts from this blog

Palm vein recognition technology  is one of the  bio metric technologies  most widely accepted by patients and healthcare providers because it identifies patients with a high level of accuracy, and is easy for patients to use and accept. While other types of biometric scanners are more popular for security systems, Vascular scanners are growing in popularity. Fingerprint scanners are more frequently used, but Naito says they generally do not provide enough data points for critical verification decisions. Since fingerprint scanners require direct contact of the finger with the scanner, dry or abraded skin can interfere with the reliability of the system. Bio metrics is gaining more and more popularity in the financial services industry worldwide. In fact, the bio metrics market is expected to reach a value of  $30 billion  by  2021 . The technology is claimed to be the most convenient method as users don’t have to remember the numbers, co...
Post a Comment
Read more

UC Browser Reportedly Under Government Scanner for Sending Data to China

UC Browser, the most popular mobile browser in India, is reportedly being probed by the government for sending data from Indian users out of the country. According to  Business Standard , "a government lab in Hyderabad is probing how Alibaba-owned UC Browser can send user details and location data to a remote server". According to reports, UC Browser sends data including device's IMEI number and location data to a server in China on connecting to a Wi-Fi network. In May 2015, a a Canadian technology research group  reported  that Alibaba Group Holding Ltd paid more than $1 billion for leaking sensitive user data and said that is a privacy risk. Citizen Lab said Chinese and English-language versions of UC Browser made easily available to third parties personally identifiable information like location, search details and mobile subscriber and device numbers. Last week, the government also  asked  all smartphone companies, majority of them Chine...
Post a Comment
Read more