The pioneering adventures of former Shanda executives: Cloud-aware founder from behind the scenes to the podium

Intermediary transaction SEO diagnosis Taobao guest Cloud host technology Hall

Tencent Science and Technology Zong December 20 Report

Star start-ups in the field of speech recognition are experiencing a covert high-level restructuring.

Tencent technology exclusively learned that Cloud CEO Liang no longer as CEO, instead focus on the technical field. Company CEO by Huang, officially took office more than a week.

The official website of the Cloud knows the truth of the news. Cloud sound in the company's internal press release bulletin, Yun-AE won the most investment value enterprises in China's top 50 title, Yun-Huang CEO attended the conference and received the award, press release and Huang to accept the award photos.

In the eyes of outsiders, this seems to be an airborne adjustment. In the speech recognition industry insiders, it is the voice recognition industry's core characters return.

"Huang finally returned, finally from behind the scenes toward the front." A voice recognition industry personage such feeling.

Who's Huang? As a leader in the field of speech recognition, he first worked in Motorola and Grand Innovation Institute, established a Grand Innovation Institute voice branch, after the grand overall strategy adjustment, he quit in 2012 to choose a business.

The argument for Huang Entrepreneurship is mixed. Grand Innovation Institute staff said, Huang is a grand innovation courtyard senior executives left the Innovation Institute, he went to set up a cloud to know sound. But in the past, Yun-ae officially denied Huang in the team.

This October, the cloud known sound confirmed the completion of Qiming venture a round of financing, the amount of 100 million yuan, at this time, Cloud know sound company was founded just over 500 days. Cloud to know the sound of the rapid leap Red also aroused the concern of the industry, but also led the Huang and Yun Zhishong contact: "Huang is not obvious on the founder of the company, more of the cloud and the sound of private relations." Although there is no clear evidence of the relationship between the two, but in fact countless, both go very close. "said one person familiar with the matter.

Tencent Science and technology learned that the Huang officially as CEO, and from the grand after the suspension of the Non-compete agreement. "Cloud to know the sound of this year's rapid growth, and Huang is closely related to the behind-the-scenes." said the person.

According to the arrangement, people familiar with the matter said that Huang is more familiar with the capital side and easier to control the overall situation. Liang is a typical technical talent, focus on technology more suitable. "In the Grand innovation courtyard when Huang is the leader of Liang, after starting a business by Liang as CEO, Huang behind the scenes." When the conditions permit, Huang to the front desk, which is very smooth. ”

As a start-up company, Cloud knowledge of a year's growth is obvious to all. "In the field of speech recognition, Baidu has been doing for more than a year, cloud know sound has also done more than a year, but the cloud know sound recognition technology such as the accuracy of recognition has been higher than Baidu a big cut." "In speech recognition startups, cloud awareness is the best," says one voice-recognition practitioner. ”

Liang the stage success of entrepreneurship attributed to the strong core technical team and platform technology team, in a year, to achieve the listed companies need 3-5 years to build speech recognition service platform. "Over the past year, the speech recognition error rate has decreased by 60% and the recognition speed has increased by more than 3 times times." Liang, CEO of Cloud, told Tencent Technology.

On this node, in the field of speech recognition on the eve of the outbreak, the resumption of cloud sound one year of dark horse-style rise, quite a lot of meaning.

Speech recognition industry explosion

"All mobile phone manufacturers are investing in voice, developing voice technology, creating more elegant designs and integrating them into mobile phones." "said Michael Thompson, senior vice president of Nuance, America's largest voice-recognition technology company.

Two years after the introduction of Siri, Apple set up a mystery office near the Massachusetts Institute of Technology (MIT) this year to develop the Siri speech recognition technology. Microsoft is developing its own voice personal assistant software code-named "Cortana" and plans to launch the next Windows Phone platform upgrade to counter Google Now and Siri. So far, Apple, Google (Weibo), Microsoft, Intel and other international giants have invested heavily in speech recognition technology.

In the Chinese market, the voice recognition technology company also has several major factions force:

The first faction from the internet giant in the voice recognition technology layout, such as Tencent, Baidu, Sogou and other products around their own ecosystem to do voice technology, used for product services. The application of micro-letter speech intercom has become a necessary part of people's Daily communication. 360 also intends to enter the field of speech recognition, and the hkust has heard about cooperation matters, but so far no clear information.

The second faction comes from the professional speech semantic recognition company on the mobile internet extension. For example, the HKUST launched a mobile internet division last year to do mobile applications, the United States nuance in Shanghai also set up offices to expand the use of Multilingual voice recognition application of the Chinese market.

The third faction comes from the emerging start-up companies, such as cloud knowledge, thinking, and so on, with the Internet company's genes, growing ferocious. There are a number of applications that specialize in speech semantics, for example, the wormhole that specializes in semantic analysis (supported by Microsoft's first Cloud Accelerator project), Smart 360 (Zhou (micro-blog) to do Angel investment), as well as to do micro-letter voice life search out to ask (the former Google Voice recognition technology staff Li Zhifei founder), They do their own part of the application of speech semantic recognition, and the other part of the technology will be based on the technology of Hkust and other vendors to complete.

Over the past few months, the wrestling based on speech recognition has entered an unprecedented white-hot phase. September 7, entrepreneurship announces the direction of redefining the human-computer interaction experience, with a high profile of man-machine conversation based on speech recognition and semantic understanding. In mid-September, the established voice technology company Hkust announced the independent research and development of the off-line voice dictation engine will be officially released in late September, will be applied to the "flying Input Method" and other products. On the October 19, the cloud informed the financing message, also released the off-line voice dictation technology, and published a semantic cloud. October 28 This year, the Hkust Flying Voice Cloud released three years, the message of the voice of the user more than 100 million.

The story of Huang

In this round of speech recognition technology boom, a figure should be the leader but lonely scattered absent, that is grand.

Compared with the hkust, the speech recognition technology of the grand year is not weak. The core character of the once-leading speech recognition team was Huang. Like Liu Qingfeng, Huang graduated from China University of Science and Technology and joined the Motorola China Research Center (MCRC) as a senior researcher in 2004. During Motorola's work, he led the world's first mobile phone voice certification system, and completed a number of audio-interactive product research and development. But during the financial crisis, motorcycles sold the entire speech recognition team to nuance.

Huang refused to be nuance incorporation, in July 2009 joined the Royal Network of Innovation Institute, October 2010 created a voice branch, trying to voice recognition technology and Shanda's interactive entertainment system, extended to the client side.

2010, The Grand Voice team in the United States National Standards and Technology Agency (NIST) held in the Speech recognition Evaluation (SRE) competition, the University of Massachusetts, Stanford Research Center, IBM and many other prestigious, famous enterprises, in 9 individual tasks to obtain 5 single first, the overall comprehensive indicators first.

Liang graduated from China University of Science and Technology, then entered the Academy of Sciences automation work, 2011 joined the Grand Innovation Institute, in the Grand Voice branch work, as senior researcher. However, with the 2012 Grand Innovation Institute Strategic Adjustment, Voice team from the Innovation Institute stripped to the Grand masters of Technology, by Danian in charge. Members of the voice team began to find their way out.

The Grand Voice team of several people chose to start a business, also do speech recognition, named "Cloud Know Sound". But the cloud is more inclined to emphasize their own technical accumulation from the Chinese Academy of Sciences automation, rather than have worked before the grand.

The role of Huang in it is intriguing.

In the summer of 2012, the staff of the Grand Innovation Hospital QQ Group, came the Huang founder Cloud know the sound of the news. A former Grand Innovation Institute staff said, "Huang is the first senior executive of the Grand Innovation Institute to leave the Innovation Institute, he went to set up the cloud to know the sound." ”

But the cloud knows the sound official denies Huang in the cloud knows the sound team.

Official information from the cloud revealed that the company's founder has two, one is Liang, as the company's CEO, the other is Kangheng, as CTO of the company, responsible for the platform Business Department.

For the relationship between Huang and the company, Tencent Technology has to the cloud to know the sound market leader to verify the matter. "Huang has nothing to do with our company," the official said. "But he also said that Huang and Liang are brother and sister relationship, will also guide the cloud to know the sound of business."

Cloud insider revealed that Huang left the Grand Institute after the establishment of mobile Internet application music radar, and cloud acoustic team in the same building office. Website information shows that two companies are in Beijing Haidian District, the Financial Wisdom International Building C block, cloud known sound in the 15 layer, music radar on 19 floors.

The head of a music radar partner confirmed that Huang was one of the founders of the music radar and talked to Huang about musical cooperation. For the suspicion of the relationship between Huang and cloud, he said, "not to create a company, but may be the way to operate the company, which in the circle is relatively normal." ”

The rhythm of the Internet

For the cloud to know the sound, in the past year, to fast win, the fastest-breaking rhythm in the field of speech recognition rapid expansion of business, to create visibility. In the same way as other companies ' voice technology solutions, cloud awareness takes the strategy of first seizing the market and then optimizing.

Last November, Cloud know Sound and Sogou voice assistant to reach cooperation, this March, and hammer science and technology to reach cooperation, May and video TV to reach cooperation, August and Inwatch, easy letter reached cooperation. Cloud know sound and the hkust to join together in the millet, Lenovo, Intelligent 360 voice assistant and many other partners voice recognition program.

Liang revealed that from the team to meet with Sogou to the first product model out of only two weeks, and this time other companies may not even talk about business terms. The collaboration is also similar, from the first and the video team to contact the internal system ran to the release in the conference only 1 months.

There is also a cooperative customer-hammer technology. CEO Luo a famous critic. This year, the week before the Hammer ROM conference, Mr. Luo complained about the delay in speech recognition, which was the CTO of the company, which has been docking the product integration and Hammer technology team. Eventually, cloud-aware and hkust flew into the hammer ROM voice solution.

Unlike previous research, Liang found that, for a year, the team has been wrestling with the technical aspects of things, he believes that with the actual business is more critical and more important. "It is also a good technology, failure is also a good technique."

Since the public cloud of speech recognition was released last year, 1000 developers have joined the platform. Through the open platform, "the threshold can be lowered enough, developers do not need to understand the specific speech semantics technology, only need to invoke Yun Zhishong services, can do application innovation." ”

Behind the open platform lies the logic of the business model of cloud-knowing: through the Voice semantic platform, the app Unicom, the user data will also be concentrated on the platform, linking each link to realize the commercial value of advertising.

Liang it as a similar Google AdSense model, although the amount of each home is very small, but through aggregation can generate commercial interests, the participants can be divided into.

The entrepreneurial rhythm of the internet to let cloud know sound tasted the sweetness, then there are risks and challenges.

One is the risk from the platform. "Tencent, Baidu do speech recognition is centered around their main design, in the ecosystem to do, but also to open the platform for the construction." "The voice-recognition practitioners argue that start-ups are riskier than other companies that already have a stable business model in speech-recognition applications."

For cloud awareness, another worry is that there are not yet core mobile internet products. At present, cloud know sound try to do voice input method plug-in, but the main business is from speech recognition solution cut, take the technology-driven route. "Start-up companies must have their own products and services, no products, it is difficult to carry user data, business interests mining more difficult." The