Please give me a safe voice assistant!
This sentence seems a bit like a whining. However, with the increasing popularity of voice assistants and smart speakers, this seems to be an urgent need.
In the first two days, I saw a fun thing on the vibrato. A beautiful woman used her male colleague’s mobile phone to confess like another female colleague. At the beginning, I also took out my mobile phone and found a female colleague to experiment. The result is true.
Although this is a funny video, it also reveals some questions worthy of our reflection:
My voice assistant, is it really safe?
I still remember seeing in the Xiaomi community that there is such a thing:
A user posts in the Xiaomi community for help. "Small love is safe when I am not at home. What should I do if I am executed by someone else?" Don't think this alarmist, but if you have Xiaomi audio, you should encounter this. The situation is obviously his own sound, but it does not know his master.
So some people jokingly said that it is not necessary to bring a mask after the crime, because he can say: XXX, please help me close all the cameras.
In fact, people who have used the Xiaomi speaker may know that the Xiaomi speaker has done a good job in speech analysis, but it is still blank in voiceprint recognition. In other words: Nowadays, many voice assistants, smart speakers and various voice products can already hear and understand what is said, but he does not know who is his master, so anyone can give orders to him.
Of course, there are two techniques involved here: one is speech recognition and the other is voiceprint recognition.
Speech recognition is better understood, and we apply more. The machine converts the speech signal into corresponding text or commands through an identification and understanding process. The most common is that WeChat can send a voice directly into text. In addition to our common social software applications, human-computer interaction, navigation, retrieval, and Tencent simultaneous transmission of the car a few days ago, this will involve speech recognition technology.
Voiceprint recognition is actually not a new concept. Everyone has more or less contact. If I change a face, face or fingerprint, then you feel that you understand a lot. At this stage, based on face, other kinds of identity authentication have been widely used. Such as Apple mobile phones, bank withdrawals, etc. (But I have to vomit, why do I apply a mask and open my phone? Is this really accurate or technical, really a vague definition?) And voiceprint recognition, in fact, with the face Like recognition or fingerprint recognition, it is a kind of biometric recognition, and there are many scenes that can be applied.
Moreover, this is a market that is currently underestimated, but as the country develops its intelligent voice platform, more and more companies and technical teams will pay attention to this field. Voiceprint recognition In all biometrics, user acceptance is the highest, and ease of use, accuracy is not lost to face recognition, and even better in identity verification. Some people have done experiments in China. They have found 23 pairs of twins, and they are the voice data of identical twins. Experts listen to them, and the resolution accuracy is about 85%. The voiceprint test by machine is 100% separated.
So, the above said so much, we can sum up in one sentence: speech recognition solves the problem you are talking about, and voiceprint recognition solves the problem of who you are.
So what are the application scenarios for voiceprint recognition?
1. Identification
In fact, with the development of the Internet, we may all feel that the traditional digital passwords are not confidential and secure, and many complicated passwords are difficult to remember. Especially with the increase of platforms, non-authentication becomes more and more cumbersome. The use of "biometric features" for verification is a good solution. Because each person's fingerprints, face, pupils, sounds and other characteristics are different, unique, the security of "living passwords" will undoubtedly be high. A lot. The voiceprints stand out from the numerous alternatives in terms of interactive nature, content changes and low cost of use, and gradually become a "black horse" in the field of identity recognition.
2. Mobile payment
In fact, many of you have been exposed to this usage scenario.
Now some financial apps have started voiceprint recognition. In addition, if you have a Tmall Elf, if you have purchased something on it, do you think that voice control is in the future?
It is worth mentioning that the Tmall Elf has added voiceprint recognition technology, which means that only you can control the Tmall Elf, don't worry about someone controlling your baby by voice, and let the wallet be lost.
3. Industrial warning
The good operation of industrial machines is critical to a business. Simply put, if we can determine machine problems before the machine completely fails, we can minimize corporate losses.
Take the factory engine as an example. The sound that is emitted when the machine is running normally is inconsistent with the sound that is emitted when it is faulty. According to the surrounding environment, the fixed sound is extracted, and the collected sound is intelligently analyzed to extract the industrial machinery. Sound characteristics, through the sound big data analysis to identify whether the machine is running properly, and based on the sound frequency, speculate the machine's operating conditions.
In fact, there are still many scenes that can be applied to voiceprints, such as electronic cases, access control of smart cars, and so on. With the popularization and landing of technology, more and more application scenarios have been expanded and developed. I believe that in the near future, we can see more use scenarios.
So what are the professional companies in China that do voiceprints?
Note that what I am talking about here is the voiceprint company, not the voice company. For example, the University of Science and Technology has a lot of research on voiceprints, but he was indeed a voice technology company. Similarly, Spirent, more are biased towards semantic analysis, speech recognition, speech understanding and other technologies.
Similar to Beijing Junlin Technology Co., Ltd., providing the world's leading audio field solutions and professional services, and promoting artificial intelligence technology products based on acoustics. The company aims to create an audio hardware platform and system platform in the field of artificial intelligence, making Junlin technology an indispensable link in the field of artificial intelligence audio. Based on this, it covers the technical chain of acoustic technology-firmware algorithm-intelligent hardware-system driver-cloud service-big data-Ai technology. Junlin Technology has a complete set of sound processing system solutions. Users can use existing or customized audio data collection solutions to upload to the cloud through the Junlin Technology Voice REST API, which is modeled, identified and certified by artificial intelligence algorithms to effectively create maximum value for the enterprise. The company has core technologies such as acoustics, wireless audio, AI acoustics, voiceprint technology, cloud and big data, product solution design conference, automotive, social security, public security, intelligent hardware and other fields.
Smart Body Fat Scales,Bathroom Scale,Bluetooth Digital Scale,Scales For Body Weight
C&Q Technology (Guangzhou) Co.,Ltd. , https://www.gzcqteq.com