Internet giant Baidu, operator of the largest search engine in China, plans to launch a new tool powered by artificial intelligence (AI) that it claims can translate languages in real time.
Its new simultaneous translation system will be showcased next week at the company’s annual Baidu World tech conference in Beijing, which starts on 1st November.
The news comes a year after Google unveiled Pixel Buds, a wireless earbud system that claims to be able to translate 40 languages in close to real time.
But the difference, according to the South China Morning Post, is that the Baidu device is to be the first with “anticipation capabilities and controllable latency”.
In other words, it is able to complete near real-time translations from languages that have very different sentence structures and can predict what subjects a speaker is about to discuss – in much the same way that a human translator prepares.
“We tackled this challenge using an idea inspired by human simultaneous interpreters,” said Liang Huang, principal scientist at Baidu, who explained that just as human interpreters familiarise themselves with a topic and speaker’s style in advance, the AI model is trained with data including similar sentence structures.
Huang added that the AI model is not intended to replace human interpreters, but instead help to reduce their workload and make simultaneous translation more affordable.
The development is being seen as a considerable breakthrough for Baidu, which recently became the first Chinese company to join the Partnership on AI (PAI), a US-based group that investigates the ethics of AI and includes Amazon, Google and other leading tech companies among its membership.
As reported by CNBC, tech firms regard digital translation as a key technology that can be integrated across a number of products, especially voice assistants.
However, Huang said Baidu isn’t planning on using it for voice assistants, and will concentrate instead on integrating it into its Wi-Fi translator. “We are still working hard on making it available to general public but we don’t know when that will happen,” he added.
Its new simultaneous translation system will be showcased next week at the company’s annual Baidu World tech conference in Beijing, which starts on 1st November.
The news comes a year after Google unveiled Pixel Buds, a wireless earbud system that claims to be able to translate 40 languages in close to real time.
But the difference, according to the South China Morning Post, is that the Baidu device is to be the first with “anticipation capabilities and controllable latency”.
In other words, it is able to complete near real-time translations from languages that have very different sentence structures and can predict what subjects a speaker is about to discuss – in much the same way that a human translator prepares.
“We tackled this challenge using an idea inspired by human simultaneous interpreters,” said Liang Huang, principal scientist at Baidu, who explained that just as human interpreters familiarise themselves with a topic and speaker’s style in advance, the AI model is trained with data including similar sentence structures.
Huang added that the AI model is not intended to replace human interpreters, but instead help to reduce their workload and make simultaneous translation more affordable.
The development is being seen as a considerable breakthrough for Baidu, which recently became the first Chinese company to join the Partnership on AI (PAI), a US-based group that investigates the ethics of AI and includes Amazon, Google and other leading tech companies among its membership.
As reported by CNBC, tech firms regard digital translation as a key technology that can be integrated across a number of products, especially voice assistants.
However, Huang said Baidu isn’t planning on using it for voice assistants, and will concentrate instead on integrating it into its Wi-Fi translator. “We are still working hard on making it available to general public but we don’t know when that will happen,” he added.
Sourced from South China Morning Post, CNBC; additional content by WARC staff
