[Xinzhiyuan Guide] Last Sunday, the "2020 China Voice Innovation and Development Summit Forum and China Voice Industry Alliance Annual Conference" was held in Tianjin. At the meeting, the ten major viewpoints of the "White Paper on China's Intelligent Voice Industry Development" were released. It fully shows us the current situation and future of the voice industry.
Remember the "two horses" dialogue at the Shanghai World Artificial Intelligence Conference last year?
Jack Ma and Elon Musk argued about artificial intelligence, one is more inclined to believe in humans, and the other believes that machines are far smarter than humans.
Who is right and wrong, and which station has the upper hand, needs to be checked by time, but what is interesting is that when this English conversation was in progress, the big screen behind them showed that AI was performing real-time speech recognition and Chinese-English translation for them.
Even if it becomes the protagonist of the discussion, AI is still functioning in an orderly manner.
This technology is provided by HKUST Xunfei. Whether AI will replace humans is still unknown, but there is a clear answer to whether AI will be applied.
On the afternoon of December 27th, the " 2020 China Voice Innovation and Development Summit Forum and China Voice Industry Alliance Annual Conference " hosted by the National Industrial Information Security Development Research Center was held in Tianjin Binhai, from politics, industry, academia, research, and users. More than a hundred representatives gathered together to focus on the hotspots of the intelligent voice technology industry and jointly plan the development of the voice industry to help the deep integration of artificial intelligence and the real economy.
The China Voice Industry Alliance aims to further integrate industrial resources, build a healthy industrial ecosystem, and promote the rapid development of the voice and language industry. In the context of Sino-US technological competition, Chinese intelligent voice companies insist on technological innovation at the source, independent and controllable technology, and effectively prevent the risk of "stuck neck".
At the meeting, the summary of the ten major viewpoints of the "White Paper on China's Intelligent Voice Industry Development" released by the National Industrial Information Security Development Research Center gave us a comprehensive display of the current situation and future of the voice industry.
The top ten viewpoints of "White Paper" are released! HKUST iFLYTEK's market share ranks first in China
The "White Paper" summary of the top ten viewpoints pointed out that the global intelligent voice and artificial intelligence industry is in the ascendant, entering large-scale development and maintaining a rapid development trend. With the support of the four basic technologies of deep learning, cloud computing, big data and 5G, intelligent voice technology and applications continue to extend to the artificial intelligence industry.
Chart: 2014-2019 Global Intelligent Voice Industry Scale and Growth Rate
It is predicted that the global intelligent voice and artificial intelligence market will exceed 20 billion U.S. dollars in 2020, and leading companies such as Google, Apple, Microsoft, and iFlytek will occupy more than 80% of the market. The global intelligent voice head enterprises have obvious advantages, continue to make efforts to construct the industrial ecology, and have huge market development potential.
Chart: 2018-2019 Global Intelligent Voice Market Share
The smart voice technologies of Google, Microsoft and Apple are mostly embodied in their C-end consumer products such as voice assistants, while iFLYTEK is in-depth in education, medical, home, office and other major application scenarios.
According to the IDC report, in the speech semantic market, HKUST IFLYTEK has the largest market share in China, followed by Baidu, Alibaba Cloud, and Spitz.
Source: IDC China, 2020
At the same time, according to iiMedia Consulting's "Special Research Report on China's Artificial Intelligence Industry in the First Half of 2020", in the smart voice track, the comprehensive strength and growth ability of HKUST iFLYTEK far surpasses competitors in the same industry, occupying the first echelon.
Figure: China's intelligent voice industry echelon, source: iiMedia Consulting
It can be seen that in terms of market share and overall strength, HKUST IFLYTEK has occupied the leading position in the domestic voice market, and the second-tier companies still have a large gap with them.
The "platform + track" bipolar amplification effect of intelligent voice is highlighted, accelerating its implementation around industry scenarios and personal scenarios. "Ripple effect" makes the current smart voice industry landing from a typical case of just-needed scenarios to large-scale applications. It has been deeply integrated in education, medical, government, finance, operators, justice and other fields; personal scenarios focus on home, car, office, etc. As the scene unfolds, consumer smart voice products continue to emerge and are frequently recognized by the market.
At the same time, my country's smart voice applications have spawned new consumption, new applications, and new markets. Especially after the outbreak of the new crown epidemic, intelligent voice applications such as smart medical assistants, telephone follow-ups, air classrooms, virtual meetings, virtual anchors, etc., helped prevent and control the epidemic and gave birth to a new blue ocean market.
Judging from the development in recent years, intelligent voice technology and industrialization are relatively mature, intelligent voice technology innovation has increased, and semantic understanding technology has made major breakthroughs.
A major event in the NLP circle this year is the release of GPT-3, which caused a huge sensation with its 175 billion parameters. You know, the human brain has only 86 billion neurons. This breakthrough has made many people think that they are one step closer to taking off the "Pearl of the Hand" of NLP.
The white paper also pointed out that the key innovations for the next development of artificial intelligence are small sample learning, dynamic adaptive, transfer learning, offline computing and emotional computing, etc. Human-machine coupling will be the long-term development trend of technology.
Under the new infrastructure strategy, intelligent voice interaction will become the portal of human-computer interaction and new information technology infrastructure in the era of the Internet of Everything, which is of great significance to the development of the country to create new competitive advantages and inject new growth momentum. Under the new development pattern of mutual promotion of domestic and international double cycles, intelligent voice technology and products are the rigid needs of human communication.
To solve the "stuck neck" problem, we need to make black technology with "temperature"
On October 8th last year, according to information published on the official website of the US Department of Commerce, many Chinese companies such as Keda Xunfei were included in the list of US entities. After the enterprise is included in the list of entities, the US government can restrict the export, import, or re-export to these institutions in accordance with the Export Administration Regulations.
HKUST iFLYTEK was the first listed company on the list that did not suspend trading due to this, and the share price rose back the next day after falling that day.
Excellent technical strength is the real reason for the uprightness of HKUST.
At the meeting, Liu Qingfeng, chairman of the China Voice Industry Alliance and chairman of iFlytek, further emphasized the strategic position of intelligent voice:
"In the era of the Internet of Everything, voice is a battleground for core technology and a key technology to prevent neck jams."
At the same time, Liu Qingfeng proposed three changes in the development of the voice industry in 2020:
1. The characteristics of application scenarios from the mobile Internet to the Internet of Everything are becoming more and more obvious, and voice is increasingly becoming the entrance to the Internet of Everything;
2. With the deepening of industry applications, the combination of voice technology and cognitive intelligence is getting closer and closer;
3. Under the guidance of the society, the development of multilingual technology is getting faster and faster.
The large-scale implementation of voice technology and products is no easy task. Large-scale production and use are quite challenging, but the 2020 epidemic has given HKUST IFLYTEK an opportunity to test its own strength.
HKUST iFlytek’s "Smart Medical Assistant" has passed the comprehensive written examination of national licensed physicians, and more than 96.3% of the candidates can quickly grasp the knowledge of new coronary pneumonia. During the epidemic, under the guidance of the National Health Commission, Xunfei used artificial intelligence to train 2 million grassroots doctors, saving medical staff a lot of valuable time.
During the lockdown period in Wuhan, through iFLYTEK’s voice technology, a single machine could make calls to hundreds of households, and one million telephone interviews could be completed within 6 hours. After the telephone interview is over, statistical reports can be generated automatically. After being promoted nationwide, 59 million visits and statistics were completed, which greatly saved follow-up time.
If this year's epidemic is a common problem facing the world, then the coming aging society is the test that my country is about to face.
Recently, the Ministry of Industry and Information Technology issued the "Special Action Plan for Internet Application Adaptation and Accessibility Improvement", which proposes to carry out the ageing and accessibility improvement of major Internet industry websites and mobile Internet applications (APP) commonly used by the elderly and the disabled.
In Tianjin, iFLYTEK is trying to create a "temperature" city.
Using voice technology, the AI system can also retrieve water, electricity and gas information. If there is an abnormality in the water and electricity information of the elderly living alone, the system will actively call to greet; if no one answers, the system will notify the family or the community, which greatly Improved social efficiency.
At this year’s 1024 Developer’s Day, there are more than 1,000 teams on the open platform of iFLYTEK, dedicated to developing applications for the deaf and blind. These applications are used by 50 million people every day, and many of them are deaf and blind. An entrepreneurial hero.
Everything behind this relies on the technology precipitation of HKUST News for many years.
Before that, Liu Qingfeng said: The basis for the redemption of AI application dividends is to break through the AI core technology gap. With a 21-year history of HKUST, iFlytek has continuously made technological breakthroughs internationally:
In the biennial International Multi-channel Speech Separation and Recognition Competition (CHiME), iFLYTEK and the National Engineering Laboratory for Speech and Language Information Processing of the University of Science and Technology of China (USTC-NELSLIP) conducted two multi-channel speech recognition at a given speaker boundary. Win the championship on the task of competition.
Since 2016, iFLYTEK has participated in this international competition for the third time and won consecutive championships. The speech recognition error rate has dropped from 46.1% of CHiME-5 to 30.5% now.
During the public ranking stage of the Global Voiceprint Recognition (VoxSRC) Contest, iFlytek also set a world record. Task 3 of the DCASE competition also won the world's first place, and the corresponding product "Industrial Stethoscope" has also been launched, which can save costs for enterprises and help industrial intelligence upgrade.
Although general artificial intelligence has just begun, the common sense reasoning of machines has not even reached the level of children. However, as long as there is enough data in the professional field, it is very worth looking forward to in the future to greatly improve the quality of life and increase efficiency through the voice system.
The technology and application of voice are related to the future of the industry, and the development of the voice industry is also inseparable from ecological construction.
Liu Qingfeng emphasized: "Only by opening up can there be endless prosperity. Ecological prosperity determines industrial prosperity." In the past year, iFLYTEK’s open platform achieved a growth of over 500,000 and the total number of developers reached 1.69 million. Among them, Tianjin "Northern Sound Valley" opened in May 2018, the developer team has reached 23,000, an increase of about 3.3 times.
As Liu Qingfeng said:
"The Chinese voice must be the best in the world by the Chinese, and the Chinese voice industry must be controlled by the Chinese themselves."
I believe that batches of artificial intelligence companies will contribute more AI wisdom to the "China Strategy" under globalization in the collaborative linkage and innovation drive, use intelligent voice to convey Chinese voice to the world, and use artificial intelligence to tell the Chinese story.