Application of various biometric technology is increasingly wide in the large environment of artificial intelligence abnormalities. Voice identification technology is a very important way of identification in the field of biological identification. Currently, there are many fields such as financial, securities, criminal investigation, and other civil safety certification systems: using sound corrugating technology, intelligent hardware products can be implemented. "Smell people"; public security organs can contain more effectively and combat crimes; financial institutions can significantly improve risk prevention system security.
In order to let users understand the sound identification technology, Next IDEA (Tencent Innovation Competition) joint Tencent Technology and the world's first security contest for smart life, Geekpwn, the first time, the first launch of using sound pattern identification technology H5 - "AI Eye Challenge", and provides the original sound material of the glory hero, let the user choose to imitate. "AI Extract Challenge" is collected by soundprints uploaded by users and analyzes the similarities.
According to the British "Guardian" report, the sound pattern recognition system can monitor more than hundreds of unique sound characteristics by analyzing a person's voice. These features include: rhythm, accent, pronunciation, and the shape and size of the roll, channel, nasal cavity corresponding to the sound. By imitation, our sound can be very like, but it is almost impossible to imitate more than 100 sound characteristics contained in a person's voice ID or some physiological features. There is a difference in the vocabulary map of any two people.
However, as the humanity, the "unique" sound created by the sound identification is really unable to break it? Ordinary people seem to be more insurmountable, and hackers use technical means to pass. Since AI's learning development model will lead to hackers' deliberate attacks, AI's own characteristics will also trigger "unconscious mutual attacks", which is the biggest hidden threat in the development process of AI. In this way, hackers can also use AI to initiate attacks to the sound pattern identification system.
On October 24th, the attack scene of the sound pattern recognition system will be staged at the 2017Geekpwn Safety Emirates Shanghai Station. It is reported that the provider of the sound sample of this competition is a dubbing actor who is glorious. At that time, the player combines analog learning and self-contained sound characteristics of the "attack" voice, and challenge the sound system. Can a synthetic voice can "imply" voice verification system? When hackers meet, what is unexpected surprises?
As a young innovation talent in the field of art and science and technology in Tencent Group, the "future imagination" proposition is proposed this year, it is awareness imagination designed to stimulate the youth group, more focused on social problems related to human future. . Artificial intelligence is naturally important propositions for NEXT IDEA. The next Idea has jointly launched the "Artificial Intelligence Safety Challenge" special for the world's first security contest, and added "AI Eyex" Scenery. The world's top hacker and artificial intelligence of the ultimate PK will be on October 24th, on November 13, in Shanghai and Silicon Valley. Technology area
Ranked between the Global R & D investment company, Samsung invested 12.2 billion euros # 4
Sharing bicycle into the Three Kingdoms kill, Yonganhang split Mobai and small yellow car
Microsoft, Fujitsu, Cisco and other 20 technology faucets jointly build the city of safe exchange data
Bitcoin price crash is listed as one of the major risks in 2018
Apple is highlight, Ar / Ai / wireless charging is included in the capsule
Our other product:
