Recently, the Spring Festival sci-fi movie "The Wandering Earth 2" is currently in theaters. The magnificent scenes, thrilling stories, and endless "hard technology" elements in the film make the audience very addicted. Among them, Zhou Zhezhi, played by teacher Li Xuejian, even contributed a textbook-level performance. Accompanied by praise, it is bound to be questioned. As early as in the trailer, many netizens expressed concern about Mr. Li's hoarse and slightly broken voice after undergoing nasopharyngeal cancer treatment. Finally, the crew used advanced AI technology to learn Mr. Li's voice print, intonation and other content through the AI model, and imitated and restored the voice in the film as much as possible before he fell ill. This created the famous scene of the "human femur" speech that we see today!
"Simulating sound with AI" sounds fantastic, but in fact this technology is gradually maturing. After adulthood, a person's voice can remain relatively stable for a long time. Each person's short-term spectrum characteristics, sound source characteristics, temporal dynamic characteristics, prosodic characteristics, and linguistic characteristics are different when speaking. Therefore, voiceprints are unique and unique. The characteristic of stability is one of the most representative characteristics of human beings. It is not only widely used in industries such as film, finance, and home furnishing, but also becomes one of the important means to solve criminal cases.
In the field of artificial intelligence and machine learning, voiceprint recognition also plays an important role. Recently, Xiamen Kuaishangtong Technology Co., Ltd. established three AI speech laboratories, the Chinese University of Hong Kong-Shenzhen Speech and Semantic Big Data Lab, the National University of Singapore-Human Speech Technology Lab, and the Southwest University of Political Science and Law-Judicial Speech Lab. On the basis of the integration of data and algorithms, the Kuaishangtong AI PaaS open platform continuously improves the basic AI capabilities of the Kuaishangtong AI PaaS open platform, enriches the technology application scenarios, and is widely used in the business of the whole industry, directly hitting the pain points of enterprise AI applications , to meet the needs of enterprises for efficient, safe and stable AI applications, to provide digital infrastructure for the entire industry, and to develop a variety of new artificial intelligence technologies, including using voice and voiceprint as the key to identify accurate identities, creating "electronic signatures" ", and apply voiceprint recognition to many fields such as finance, public security, home furnishing, and automobiles.
Compared with face, fingerprint and iris recognition, voiceprint recognition has many advantages. First of all, the voiceprint corpus is collected in a more natural way and is not restricted by specific scenarios, and the voiceprint is non-contact, suitable for remote operation, low cost and convenient. Compared with other recognition methods, voiceprint recognition has a higher safety factor, and is suitable for various scenarios such as smart home, security protection, and criminal investigation. From the perspective of development prospects, the voiceprint recognition market has huge potential and promising prospects. As early as a few years ago, Kuaishangtong had already laid out the voiceprint technology market in advance, and invited Professor Li Haizhou, one of the earliest scholars in the world who began to study voiceprint technology, and an academician of the Singapore Academy of Engineering, to become the chief scientist of Kuaishangtong to jointly tackle difficulties and promote domestic Voiceprint technology development and application.
At present, based on the mature AI basic capability construction of the AI PaaS open platform, Kuaishangtong has mastered the voiceprint recognition technology proficiently, and has rich experience in voice characteristics and models. The Chinese voice recognition rate is far ahead, and the voice voice AI technologies such as fingerprint recognition and semantic analysis have been put into use in finance, public security and law, and have broad prospects in smart home, smart education, social security verification, and telecommunications fraud prevention.
For example, in recent years, telecommunications fraud has shown a blowout trend, and it is not uncommon for criminals to defraud money by disguising themselves. Kuaishangtong uses voiceprint recognition technology to create a variety of AI voiceprint tools such as "on-site voiceprint rapid extraction and comparison system" and "VoiceSense intelligent voiceprint identification expert system" to assist law enforcement and provide public security agencies with the ability to detect telecom fraud cases. assist. And it stood out among the 246 reported projects in the second national criminal technology "Double Ten Plan" tackling key innovation competition, and won one gold and one bronze. Ministry of Police Equipment Procurement Catalog.
In today's era of "Internet of Everything", voice and voiceprint technology can be regarded as one of the most critical entrances in this era, which will help realize language intercommunication and build a community of shared future for mankind. In the future, Kuaishangtong will better integrate the development of "industry-university-research integration", continue to promote the technological innovation of the underlying AI capabilities, and realize large-scale application landing. Better use artificial intelligence technology to serve the society and build a better world, help China's artificial intelligence win the right to speak in the world, realize more innovative applications of artificial intelligence, and truly solve social needs.
About Kuaishangtong
Kuaishangtong is an officially certified future "unicorn" enterprise in the field of digital economy. It is a leading private enterprise in the field of artificial intelligence in Fujian Province. It has completely independent intellectual property rights. Its core technical team is led by internationally renowned academicians. The industry has a wide range of influence. Its self-developed "Kaishangtong AI SaaS Enterprise Service Platform" is used by more than 380,000 small and medium-sized enterprises around the world, and has mastered a number of world-leading technologies, especially natural language processing, voiceprint recognition and other technologies. Cooperative customers include Guizhou Provincial Public Security Department, Bank of Communications, etc.
