Dialogue with Spies: The era of voice + AIGC has just begun

The value of Aspire lies in creating a new chapter in building intelligent software and hardware with human-computer interaction solutions based on voice technology. Concretely speaking, it is not only ChatBot, but more like "Jarvis".

@商科星球original

Author丨Yuan Jing

Edit丨Big Rabbit

Halfway through 2023, artificial intelligence has become the focus of attention of the whole society. Today, we are making great strides towards the era of intelligence.

In the past, the team of Digital Planet (ID: digital-planet) has visited nearly a hundred related companies. Some companies have demonstrated the ability to implement their applications, and Spies is one of them.

Founded in 2007, Aspire can be regarded as a senior enterprise in the artificial intelligence industry. Both graduates of Cambridge University, Gao Shixing and Yu Kai have laid a full technical brand for this company. Now Spies has become the leading artificial intelligence enterprise in Suzhou, and its products provide artificial intelligence solutions for different industries, including speech recognition, natural language processing, machine learning and other technologies.

A few days ago, at the DFM-2 Dongfeng large-scale model launch conference of Spiech, the founder and chief scientist of the company demonstrated the strategy and tactics of the company in detail. Compared with other industry players, Aspire is more pragmatic and far-sighted.

01

A “special purpose model” for the service industry

Leaving aside the big model, Aspire is a human-computer interaction solution provider with AI speech language technology as the core.

For example, in remote conference scenarios, it has a noise-canceling microphone speaker, which combines functions such as 360° omnidirectional sound recognition, voice-to-text conversion, meeting records, and Bluetooth speakers. This series of products can effectively filter the surrounding noise and record the speech content of multiple people. In the "directional sound recognition" function, you can specify and recognize the speaking area and automatically eliminate noise in other areas.

Such software and hardware AI products can also be applied to scenarios such as smart homes, smart customer service, and smart vehicles, and are widely used in the fields of finance, transportation, and government affairs.

After the advent of the large-scale model era, the DFM-2 large-scale model developed by Spiech came out, and AGI began to apply to specific industries. In this competition around "true artificial intelligence", Aspire has two major advantages in voice technology and landing ability. The concept of "special model" promoted by it is different from general-purpose large-scale model products, and aims to focus more on landing in specific industries and promote intelligent upgrades in different industries.

At present, the DFM-2 large model has more than 10 billion parameters.

02

Intelligent Prospects of Vertical Industries

Voice combined with large models, more like "Jarvis" in "Iron Man".

This is a technical route that looks fancy but is difficult to do. However, what makes Apex confident is that they have accumulated more than ten years in a series of technologies including front-end acoustic signal processing, speech recognition, voiceprint recognition, natural language understanding and generation, dialogue management and speech synthesis.

Dare to be different this time, in fact, it is a long-term accumulation.

Although it is intelligent, in the scenarios of Internet of Vehicles, Internet of Things, finance, public security and government affairs, different fields have different needs for intelligence. Therefore, their requirements in terms of hardware power consumption, computing power, security, cloud configuration, and cross-system operation are also different.

In order to achieve a better user experience and enable intelligent products to be used in different industries, the solution given by Aspire is to build a dedicated model and use standardized solutions to help it build a full-link man-machine dialogue capability. "To this end, we have launched a variety of standardized AI products from software to hardware, including dialogue software, chips, modules and complete machine products." planet) said.

Obviously, this method can better enable large models to realize the data-capability cycle, and at the same time, it also poses a lot of challenges to intelligent suppliers.

Gao Shixing believes that in the future, every industry may have a super brain or super assistant, and in this process, the intelligent incremental scene will be the first to eat the first share.

For example, in the automobile or electric vehicle industry, human-machine dialogue will be embedded in the intelligent system to further activate the user's voice interaction needs in the "third space";

As the living room, bedroom, kitchen, bathroom, balcony, sports and leisure, entertainment, office and other scenes gradually realize intelligent upgrades, home appliances and even tea bar machines, mahjong tables, massage chairs, treadmills and other products will have new intelligence. Human-machine dialogue will become the characteristic selling point of new products.

It is worth noting that the "dedicated model" of Aspire puts more emphasis on human-computer interaction solutions, and is a bridge combining natural language (especially speech) and hardware. It belongs to the industry language model with general intelligence for the vertical domain . Yu Kai, co-founder and chief scientist of the company, emphasized: "We have no competitive relationship with other vertical industry models, and we can cooperate with each other to put technology into industrial applications in their fields."

03

Complexity and Particularity in the To B Industry

As we all know, the development of artificial intelligence technology is promoting the intelligent upgrading of various industries, especially in the field of To B.

Compared with the To C field, the enterprise market has higher requirements for the accuracy and security of the generated content. This can be seen not only as a new incremental market, but also as a new opportunity for AI companies.

The industry solution given by Aspire is to gradually develop large-scale customization capabilities and product standardization capabilities based on general artificial intelligence. Based on a variety of industry solutions to cover the needs of "thousands of people and faces" of client companies. According to Yu Kai, its personalized voice interaction solution has realized the "flexible batch manufacturing" of the man-machine oral dialogue system.

Among them, in the field of automobiles, the company's products have entered the car front-loading and after-installation markets, and have obtained more than 100 patents from 26 car companies including SAIC-GM-Wuling, Great Wall, Hezhong New Energy, BYD, Xiaopeng, and Ideal. The mass production of a model "on the car". In addition, in its latest cooperation list, there are also international partners such as Mercedes-Benz.

In the home appliance industry, Aspire has won many head partners such as Haier, Hisense, Changhong, Meiling, Robam, Zoumi, Ecovacs and Yunmi. Among the new products, smart assistants and smart secretaries have become one of the functions that consumers love.

In finance, the need for voice is also growing. Taking the intelligent outbound robot as an example, Aspire already has many customers such as Bank of Communications, China Construction Bank, Shanghai Pudong Development Bank, China Merchants Bank and China CITIC Bank.

Since its establishment in 2007, it is the first in China to introduce the concept of man-machine dialogue into China, and then release the DFM-2 Dongfeng large model in 2023. In recent years, the teams of Gao Shixing, Yu Kai, and Spies have gone through the most difficult period of artificial intelligence development and are beginning to usher in the dawn. Now, the solutions launched by this company are being used in large quantities in various industries.

It is foreseeable that in the near future, every industry can make its own "Jarvis".

Guess you like

Origin blog.csdn.net/m0_73135814/article/details/131808693