The scale of the smart voice market will grow by 44% in 2021

  Break through the core technology and strengthen the voice industry (new perspective)

  "Hello, welcome to the Beijing Winter Olympics."

  As soon as the voice fell, it was converted into English, French, Japanese and other languages ​​to broadcast in sequence.

Walking into the Beijing Winter Olympics cabin, a virtual anchor named "Aiga" has officially taken up his post. With the support of intelligent speech recognition technology, it can translate Mandarin into multiple languages ​​around the world in real time, allowing Olympic sounds to spread all over the world faster .

  Not only a virtual anchor, but also an AI (artificial intelligence) personal education integrated with a dressing mirror, a smart wearable device that helps the courier quickly receive and dispatch items, and a mouse that can "understand" voice typing and surfing the Internet... Now, more More and more intelligent voice technologies are moving from the laboratory to the terminal application, entering and serving people's daily life.

  "As an important part of the software industry, the intelligent voice industry has entered a new stage of rapid development." At the China Intelligent Voice Industry Development Summit Forum held a few days ago, Wang Jianwei, deputy director of the Information Technology Development Department of the Ministry of Industry and Information Technology, introduced that in recent years, my country's intelligent voice industry The speech industry is booming, with breakthroughs in core technologies, and the current speech recognition accuracy rate has reached 98%.

  The newly released "White Paper on China's Voice Industry Development 2020-2021" (hereinafter referred to as "White Paper") shows that my country's smart voice market will reach 21.7 billion yuan in 2020, a year-on-year increase of 31%, and it can reach 28.5 billion yuan in 2021, a year-on-year increase of 44 %, to effectively drive the digital development of the industry.

"In the era of the Internet of Everything, more and more smart devices need to be controlled from a certain distance, bringing development opportunities to the smart voice industry." Liu Qingfeng, chairman of the China Voice Industry Alliance and chairman of iFLYTEK, introduced, voice interaction empowers Smart devices are growing rapidly. Take iFlytek as an example, the volume of voice assistant interaction in 2021 will increase by 84% year-on-year.

  As my country's intelligent voice industry enters a period of large-scale and deep cultivation, how to accelerate the development and industrialization of key technologies and promote the continuous expansion and strengthening of the industry has become a common concern in the industry.

  "At present, the development of intelligent voice technology is facing three major challenges: multilingual language interoperability, complex scene human-computer interaction, and multimodal virtual world." Liu Qingfeng analyzed that multilingualism not only refers to foreign languages ​​but also domestic dialects; complex scenes require high noise To achieve accurate recognition in multi-person speaking scenarios, the recognition rate of iFLYTEK’s products is expected to increase from 69% to 80% this year; multi-modal interaction is to add factors such as tone, tone, expression, and mouth shape to the voice to make the perception more intelligent.

  The "White Paper" pointed out that the key innovations for the future development of intelligent voice are unsupervised learning, multi-modal integration, and cross-integration innovation of brain science.

At present, breakthroughs are needed in algorithms such as unsupervised learning and low-resource model algorithms.

In addition, there is still a gap between my country and the international advanced level in the field of AI chips, which is the basis of computing power.

  To promote the high-quality development of the intelligent voice industry, the Ministry of Industry and Information Technology will carry out three tasks in the next step.

Wang Jianwei introduced that the first is to encourage local governments to speed up the formulation of industrial policies that are conducive to the integration of intelligent voice technology and the real economy.

The second is to encourage leading enterprises and scientific research institutions to jointly carry out technological research, further improve the technical level of speech recognition, synthesis, interaction and voice chips, and build a public service platform such as national inspection and testing to provide strong support for industrial development.

At present, the China Voice Industry Alliance has attracted more than 70 enterprises with core technologies in the upstream and downstream of the industry chain. In the future, it will add another 70 related enterprises and encourage more scientific research institutions to join the alliance.

The third is to continuously expand the application scenarios and accelerate the integrated application of voice technology in the fields of smart manufacturing, smart home, smart healthcare, education and elderly care.

  Our reporter Han Xin