The foremost Internet search provider of Chinese language, Baidu, Inc. has launched DuerOS Prometheus Project recently for a rapid progression in the conversation AI capabilities. For conversational AI, this project contains of interdisciplinary association, a $1 million capital for investment, a huge open datasets for nurturing the talent in this cosmos. Baidu has aim for creating a simpler world from a complex one for the users by enterprising them through technology.
For this project, the pronouncement event was organized in Redwood City, California. In that event, Kaihua Zhu, the CTO of Baidu’s DuerOS, has quoted that the way of communicating to the devices has up surged the voice. For the speedy expansion of conversational AI, monetary incentives, interdisciplinary associations and open datasets would form the conditions which are essential.
Dueros: An Expertise of Conversation with Its Three Bulky Datasets
Baidu’s conversational AI platform, DuerOS has conversational expertise in 10 main domains along with more than 100 sub-domains. It has become the most preferable platform for China’s third party hardware producers during the opening of its promotion in 2017. It has incorporated DuerOS in over 100 branded products ranging from air conditioners and refrigerators to storytelling machines, nifty speakers and set-top boxes of TV.
After this event, three hefty datasets in- speech recognition, wake word detection from a faraway and multi-turn discussions will steadily be unlocked by Baidu for qualifying technologists and developers. Beside these, this will also train and improve their conversational AI systems’ algorithms. In the dataset of wake word recognition, around 500,000 audio clips of five to ten famous Chinese wake words will be involved. It also comprises “xiaodu xiaodu” which is the wake word to trigger DuerOS empowered devices.
The datasets of speech recognition will contain thousands of hours for Mandarin speech recognition information. This is due to allow people for training the systems that can correctly listen human speech under any difficult condition like- loud surroundings. Along with these, this project will also discharge thousands of conversation information that will include 10 dissimilar domains for the advertisement of the advance technology for multi-turn dialogue.
Baidu’s Imminent Conversational Computing
The Principal Architect of Baidu for DuerOS, Guoguo Chen has stated that data has become the new oil in AI’s age. Beside this, for stopping individuals and several smaller organizations from emerging prominent verge of the conversational AI systems, it works as a fence. Chen also has added up that they are anticipating to hasten the innovational step in this space to develop the forthcoming of conversational computing. And, this would be possible by opening their dataset, monetary incentives and proposing interdisciplinary associations.
Project Prometheus will also fix agendas with research organizations and universities for conducting course design, workshops and joint training, in addition to the open dataset. As a result, a productive exchanges on the conversational AI will be endorsed and this will entice the best talent in this galaxy. The Baidu Duer Business Unit sponsored the DuerOS Prometheus project in collaboration with Baidu Cloud, Baidu Speech Technology Group and Baidu Campus Branding. In this way, Baidu will be able to reach its goal on the conversational AI and also with new extensions.