On July 5th, 2017, Spirit and CCF-GAIR held the workshop of “AI is DUI, Spirit Developer Developer Camp†in Shenzhen. This is the first show of SBI's DUI open platform. More than 100 veterans and outstanding developers from well-known enterprises and academic circles such as Alibaba, Baidu, 360, Foxconn, Diefang, Chinese Academy of Sciences, Shenzhen University, etc., Spirent CEO Gao Shixing, VP Lei Xiongguo, VP Zhao Hengyi, etc. The scene brought the debut of the Spirit DUI open platform.
The outbreak of artificial intelligence has led to the emergence of various types of intelligent hardware products. The application of AI technology has promoted the transformation of human-computer interaction. Intelligent voice has made the control easier, and gradually eliminated the buttons, remote control, APP and other terminals, becoming the standard of intelligent hardware. With interactive mode. However, the development and application of AI voice technology has technical thresholds. Amazon Alexa, which is popular all over the world, is not convinced in the Chinese market, which leads to more product companies can only be discouraged, thus providing more manufacturers/personal developers with high self-sufficiency. The defined open platform became just needed. As far as the AI ​​open platform is concerned, although many companies in China currently use the “Chinese version of Alexaâ€, the overall market is uneven, and it is still in the initial stage of development, which provides a huge development opportunity for AI companies that do platform in China.
Figure 1 Spirent DUI open platform first show site
According to the introduction of Spirit CMO Longmengzhu, in 2014, Spirit released the first domestic spoken dialogue system platform “Spirit Dialogue Workshopâ€, which took the first step of the platform, and created high-availability customization for the pursuit of experience upgrade. Dialogue platform, Spirit has planned the DUI open platform since the beginning of 2016 and has modularized the operation in combination with the business. At the end of 2016, the upgrade has become one of the company's major strategic directions. Compared with the use of Amazon Alexa, you need to configure AWS and credit card. It is suitable for professional programmers with certain technical capabilities. The operation threshold of SBI is very low, and the registration is ready for use. Non-professional developers can quickly get started. DUI platform It has gathered many years of speech technology accumulation and has performed particularly well in the Chinese language field.
Figure 2 Si Bi Chi VP Zhao Hengyi Interpretation of DUI Open Platform
The Spirit DUI platform was built by Spirit VP Zhao Hengyi and a team of nearly 100 people. At the workshop, he admitted that “the birth of the platform stems from the pain point. The AIOS dialogue operating system has a very wide customer base. AIOS has been applied to a variety of intelligent hardware products. In the process of polishing products with hardware manufacturers, Spirit found that many manufacturers are limited by the development of custom product functions, system iterations, content upgrades, etc. The open platform will directly address this pain point, shorten the vendor development process, and better empower smart voice to developers and enterprises."
Figure 3 Developers actual combat
Figure 4 Field combat map
The team will elaborate and demonstrate the DUI open platform, and collected 20 developers to conduct practical exercises. In just 30 minutes, the dialogue skills customization and experience were quickly completed. Throughout the process, the DUI open platform highlights several distinct advantages:
[One-stop dialogue system customization platform]
Si Bi Chi is currently the leading artificial intelligence voice technology enterprise in China. It adheres to the integration of production, education and research. With decades of voice technology accumulation, Spirit will open its core technical capabilities to developers through the DUI open platform. The platform-based dialogue system custom development platform covers speech recognition, speech synthesis, semantic understanding, intelligent dialogue, multi-round interaction, correction interruption, echo denoising, etc., providing both single technical services and complete intelligent dialogue interaction. solution.
[High availability customization]
Zhao Hengyi said that the SBI DUI open platform is the first truly ultra-highly customized platform in China. Each module can be customized, for example, GUI customization, wake-up word customization, and skill deep customization; both provide universal Scenario dialogue and content skills also support developers to fully customize the dialog logic and content, access third-party services, and provide detailed customization to each round of interaction, allowing developers to customize the most appropriate voice features based on product characteristics. In terms of speech synthesis, Spirit DUI opens more than ten kinds of personalized synthesizing sounds, supports rapid custom synthesis, and so on, to enhance the personalized features of the product.
[Cloud real-time release updates, local easy to solve no network problems]
For intelligent voice, the stable network environment in the home is a good application scenario. How to deal with the poor network environment or even the networkless environment becomes a big problem, and it also becomes the just-needed function point of the outdoor scene. The Spirit DUI open platform “cloud+end†solution is very competitive. The platform provides both cloud API and local SDK. Developers can open the platform through DUI, and can use content, engine, dialogue logic, hot words, etc. in the cloud. Real-time updates, local voice wake-up, offline recognition, offline skills, etc. support the basic use of the product in the case of poor network or even no network.
[Data visualization, multi-version management, operation and maintenance without stress]
The SBI DUI open platform provides a visual big data platform. Developers can customize real-time data by defining cycles, dimensions and scopes to help developers improve product service performance and continuously optimize user experience. At the same time, the DUI open platform will greatly reduce the developer's operation and maintenance pressure, support multi-version management, open enterprise accounts, set different responsibilities and permissions according to different roles, and promote team collaboration.
[Vertical application scene, rich skill store]
Based on the current business development of Spirit, the SUIDUI open platform mainly covers smart car, smart home, intelligent robot, story machine, mobile assistant and other scenarios. The subsequent DUI platform will gradually extend to other fields, for more The field performs scene optimization and skill realization. The professional and huge “skill shop†can cover the diverse needs of different scenarios. It supports local skills such as making calls and alarm settings, as well as third-party service resource access, and is constantly enriched and improved.
According to the site introduction, the current SBI DUI open platform is still in the internal testing stage, and has not been officially launched. More functions and highlights of DUI will be exposed in the subsequent workshop activities. Zhao Hengyi said that “Spirit has invested a lot of attention in the DUI open platform. In addition to setting up a team of 100 people, Spirit has set up a 200 million artificial intelligence industry fund to help the development of the artificial intelligence ecosystem.†The entire artificial intelligence ecological layout of the company has been fully developed.
Yann LeCun, the father of convolutional nerves, frequently entered the country in the past six months. In several conferences, it was mentioned that the future of artificial intelligence cannot be predicted, but what is certain is the future development direction of artificial intelligence. It must be that the machine has some independent thinking. Predictive ability to mimic the laws of humans and nature. Perhaps it is still too early for the machine to imitate and predict independently, but the Sibo DUI open platform allows us to find some paths: Since the machine can't think and predict independently, then we may be able to make the machine have some imitation ability.
The first show of the SBI DUI open platform has brought surprises to the industry. AI technology has promoted the advancement of the Internet of Things, intelligent hardware and other fields, and will also empower traditional industries, bringing greater value and efficiency. The Sibo DUI open platform greatly enhances the enabling speed and application efficiency of intelligent voice technology. Platform products bring the expansion of AI technology value, which may be the only way for human-computer interaction transformation and development.
It is reported that before the official release of the Spirit DUI platform, a number of closed developer workshops will be launched in Shenzhen, Beijing, Shanghai, Hangzhou, Suzhou, etc. The specific information can directly pay attention to the public number of the Spirit DUI platform.
Flum Float
Nanning Ousibang Information Technology Co., Ltd. , https://www.ousibangvape.com