Jiwei Network reported that recently, Shanghai Shencong Semiconductor Co., Ltd. (hereinafter referred to as "Shencong Intelligence") announced the completion of a round of financing of hundreds of millions of yuan, investors including Yadi Technology Group, Zhuhai Dahengqin Group, Yuanhe Holdings, Suzhou Industrial Park Science and Technology Innovation Fund, and Sibichi. S&S. Intelligence will continue to increase chip research and development, continue to give full play to the advantages of the "cloud + core" integrated strategic layout, accelerate product optimization and upgrading, and expand the overall team investment to provide customers with first-class overall solutions.
S&Sint Intelligence is a chip design enterprise under Spitzsch, relying on Spich's technology and resource empowerment, focusing on creating an integrated overall solution of "algorithm + chip". Dr. Zhou Weida, co-founder and chairman of S&S Intelligence, said that S&SInt Intelligence provides a "chip + algorithm" integration solution, using algorithms to define hardware and software-defined chips, and making products towards future needs.
At the beginning of its establishment, S&S Smart received investment from SMIC Juyuan, the SMIC International Industry Fund; at the end of 2020, it received tens of millions of yuan of Pre-A round financing; today, hundreds of millions of yuan A round of financing highlights the confidence of capital and the market in its development.
AI lands and aims at the target track
With the continuous rise in the popularity of electronic devices and the continuous expansion of Internet of Things application scenarios, it has become an industry consensus that voice will become an important interaction method for the next generation of smart devices.
According to the "2020-2021 White Paper on the Development of the Chinese Industry", the scale of China's intelligent voice market has reached 21.7 billion yuan in 2020, an increase of 31% year-on-year, while the year-on-year growth rate in 2021 can reach 44%, and the industrial prospects are very broad.
Based on the judgment of the prospect of the speech recognition market, Relying on the intelligent voice technology and resource empowerment of TheLbichi Group, Shencong Intelligent has aimed at the three major application directions of smart home, intelligent office and intelligent vehicle since its inception, deeply cultivated the scene in a combination of soft and hard, and quickly entered the market with its leading technology.
Shencong intelligent generation Taihang chip TH1520 was lit up and verified in 2019, and mass production was completed in 2020. The product has been applied in three major scenarios such as smart home appliances, smart office and smart vehicle field, and has confirmed in-depth cooperation with the head enterprises in more than 30 industries such as Midea, Hisense, Yadi, and Staring Pai; Taihang's second-generation artificial intelligence SOC chip TH2608 was unveiled in May 2021 and will be mass-produced in 2022; at the same time, the second low-power chip of Taihang II, YT2128, will also be listed synchronously in 2022, focusing on offline low-power voice control.
Since the mass production of the first chip of Shencong Intelligence in 2020, the shipment volume in 2021 has climbed rapidly and achieved several times growth. At present, Shencong Intelligence has successfully introduced many domestic first-line smart homes, smart conferences and other manufacturers. Dr. Zhou Weida said that in the dedicated voice chip market, the market share of Shencong Intelligence has exceeded 20%, and the competitiveness of the product is very obvious.
Up to now, Shencong intelligent products have won a number of heavyweight awards and successfully went to sea: for two consecutive years, they have been selected into the "AI Chip Recommendation Catalog" of the AIIA of the Ministry of Industry and Information Technology of the People's Republic of China, and passed the international SGS three-system certification. It has been selected for KPMG's "Core Technology 50 List" for two consecutive years, and a generation of chips has also been selected for the transformation of high-tech achievements. At the same time, in March 2021, the generation of Sicong Intelligent's chips also became the first voice chip in China to pass amazon Alexa certification in the United States; in September of the same year, it passed the Microsoft Teams certification.
From smart home, smart office to smart car, the commercialization of S&S intelligence continues to accelerate.
Dr. Zhou Weida said, "This financing is of great strategic significance for the further development of Sicong Intelligence. The investment party, Yadi Technology Group, will drive deeper technical collaboration and product development between the two sides, and the two sides will accelerate to become closer strategic partners. Shencong intelligent chip will empower Yadi products to upgrade intelligently, provide them with a stable and smooth intelligent voice interaction experience, and open a new intelligent era of human-computer interaction of two-wheeled electric vehicles. ”
Cloud + core integration strategy, multi-dimensional cross-border technology integration
Although the market scale is expanding rapidly, but the high investment in artificial intelligence and speech recognition system research and development characteristics, but also put forward a very high requirements for enterprise technology, and Sicong Intelligence since its inception with distinct technical characteristics, its founding team has been in SMIC, Intel, Imagination and other core executives, the team has a number of doctors and postdocs, the entire team has a wealth of semiconductor industry and artificial intelligence algorithm experience, and constantly accelerate the application of products in many fields.
Focusing on speech recognition, opening up the underlying research of products and the development of artificial intelligence, S&Satoshi Intelligence is not limited to the cloud side or the end side, but adopts the cloud + core integration strategy - Sicong intelligent AI chip + DUI platform, combining artificial intelligence technology and mature chip industry, accelerating the upgrading of the traditional chip industry, and opening up the intelligent, low-power intelligent voice chip market.
At the level of the Internet of Things to the cloud, the Subich DUI platform has opened up more than 80 smart home platforms after several years of polishing, and has been iteratively updated in the feedback of many developers. In this process, Shencong Intelligent has accelerated the commercialization of product technology through the layout of the DUI platform and diversified solutions.
SPICH voice technology services have accumulated a large number of customer needs, so that Sicong Intelligent has the advantage of accurate judgment of customer needs, and under the support of the underlying data, the accuracy of speech recognition technology has also been continuously improved, providing a strong market driving force for the commercial penetration of S&S smart products.
At the end-side level, speech signal processing technology, speech wake-up technology, recognition technology, dialogue technology, voiceprint technology, and integrated NPU deep learning chip design, making the energy efficiency ratio a hundred times higher. The in-depth cooperation with SMIC and TSMC also guarantees the advantages of Sicong Intelligent Supply Chain, in the global chip production capacity shortage environment, S&S.Smart because of its unique product competitiveness, the obtained production capacity can fully guarantee all customer needs.
Shencong Intelligent's generation chip Taihang TH1520, equipped with Si Bi Chi full-link artificial intelligence speech technology, the advantages of low-power algorithms; the subsequent launch of the Taihang SECOND generation TH2608, which integrates the functions of speech, control and connection, further iteration and optimization, and the ability of the application side is more prominent and integrates the mixed recognition capabilities of Mandarin and dialects, emotion recognition capabilities, voiceprint recognition capabilities, speech synthesis capabilities, direct face wake-up capabilities, multi-modal interaction capabilities, and has been further optimized in terms of user experience.
Through the combination of soft and hardware and multi-scene linkage, Shencong Intelligent uses products to land speech recognition into intelligent conferences, intelligent calls, smart homes, and intelligent vehicle scenarios, providing users with a faster and more convenient experience.
Speech recognition "black technology" opens up the full link of voice interaction
As the speech recognition market continues to expand, so do the number of challengers entering the field. The one-stop chip + voice solution provided by Sicong Intelligence superimposes the full-link voice algorithm and becomes one of its most important competitive advantages.
To do a complete human-machine dialogue product, the middle technology chain is very long, Sicong Intelligence with its leading algorithm advantages, through the leading voice signal processing, speech wake-up, speech recognition and original direct-faced wake-up technology, binary neural network technology, to create a full-link voice interaction key technology.
With the popularity of smart homes, how to achieve priority wake-up of multiple smart devices in the same scenario has become a problem. When devices share the same wake word, multiple smart devices may wake up at the same time. The "face-to-face wake-up" algorithm used by Shencong Intelligence, combined with device microphone data, sound source positioning and other information, when multiple devices with the same wake-up word are connected to the same LOCAL area network in the same home environment, with the communication between devices under the distributed networking strategy, the best unique device is selected for response based on the indicators of all devices, and the judgment is more accurate and more in line with the user's interaction habits, which solves the multi-device collaboration problem in the whole house intelligence.
At the signal processing level, low power consumption has become an important issue, and Sicatoshi Believes that low power consumption does not mean reduced reliability. Through the optimization of all-scenario power consumption, S&S smart can ensure the overall performance under low power consumption, and after the accumulation of technology, S&Satoshi has already had core advantages in the low-power direction of voice calls.
In addition, in the process of information processing, Sicong Intelligence's original binary neural network technology has greatly improved the efficiency of AI computing while ensuring the accuracy of calculation.
In the AI landing process, user experience as a key part, voice interaction technology accompanied by smart home and other scenarios into thousands of households, a large number of computing in the cloud will face delay and stability problems, for network latency and user privacy and other aspects of consideration, Sicong Intelligence continues to optimize offline and online solutions, the current TH1520 speech recognition chip has supported full offline recognition, can identify up to 200 instructions offline, TH2608 its offline recognition capabilities will be further strengthened.
Through the integration of algorithms, chips, software and hardware, Shencong Intelligent multi-dimensional cross-border technology integration, joint upstream and downstream ecological partners, to open up the underlying research of products and the development of artificial intelligence. Facing the future, Sicong Intelligence has also clarified the direction: planning from the co-chip to the main control chip, integrating AI, connection, master control, multi-modal and other all-in-one SoCs, so that the chip can empower everything. (Proofreading/Sami)