Last Sunday, voice solution provider Yun Zhisheng held the industry salon of "Nuclear Fusion - Yun Zhisheng Opens a New Chapter in Smart Home" in Shenzhen, invited many important guests in the smart home industry, and brought many partners' products for on-site display.
Far field speech recognition, a new universal entrance
After smart phones, routers, TV boxes, and even webcams have been used as the entrance to smart homes. The mobile phone among them is also known as the all-purpose portal. Basically, all smart devices have their own mobile apps. But although this entrance is omnipotent, is it really suitable for various devices? The development of smart home devices in the past two years has clearly given a negative answer.
Although the mobile phone is used by everyone and is used more frequently, the daily control of household appliances is simple but extremely scattered, and the mobile phone control will be troublesome. If simple operations such as weighing, turning on and off lights, and even adjusting the air conditioner can only be completed by turning on the phone and finding the app, it will be a huge setback in the experience. In addition to the daily physical keys and smart phones, the daily control of smart home needs a simple and easy access with high response speed.
And voice control naturally came into everyone's view. Voice operation like the movie Iron Man can easily complete the control operation of various devices. The whole operation process does not even require additional learning of the sci-fi scene, in fact, it is not far from our life. The technology required for the entire function is actually mature.
"Double wheat" far-field recognition technology
On the day of salon, Yun Zhisheng first disclosed the "far-field speech recognition scheme based on dual microphones", which is a technology that can translate the above scenarios into reality. The scheme adopts SSP technology, and two microphones placed in different positions are used in the hardware, which can effectively suppress the noise and reverberation effects beyond the user's voice, so that the far-field pickup can be effectively carried out in more than 95% of the scenes. With the far-field speech recognition engine of Yunzhisheng, the accurate recognition effect can be achieved within a distance of 5 meters.
Far field speech recognition technology has a better practical effect than "beamforming" technology, especially the voice and noise can be located at any position of the microphone (or even at the same angle). Moreover, since this scheme only requires two microphones, the installation position is flexible, and the equipment orientation is not considered, the overall cost is lower, the equipment structure and appearance requirements are simpler, and it has obvious advantages over beam forming technology. At the same time as the far-field recognition technology, "low-power voice wake-up technology" is also provided to provide a complete voice interaction experience.
For many mobile phone enthusiasts, "far-field recognition" and "low-power voice wake-up technology" may be quite familiar. The application of similar technologies on smart devices was first on Motorola's moto X, which has an independent all-weather standby voice processing chip, which can still maintain voice wake-up standby when the screen is turned off. Later, Apple's iPhone, Meizu's MX4 Pro, Samsung's S6/S6 edge and other flagship phones are also equipped with similar "far field recognition" and "low-power voice wake-up technology".
Make home equipment have direct interaction ability
Although similar technologies have accumulated a certain amount of technology in mobile phones, there is still no relevant product launch in smart products, and even no similar solution provider before. The "double microphone" far-field recognition technology disclosed by Cloud Zhisheng this time has been packaged into different chip schemes, and the "far-field recognition" capability can be obtained by adding two microphones to the intelligent hardware.
This solution makes it possible for users to interact directly with intelligent hardware, and provides these devices with the opportunity to work completely independently of mobile phones. The salon demonstrated the functional characteristics of the home robot equipped with this technology on the spot, as well as the measured video of voice control of related home devices.
Users can sit on the sofa and directly control the air conditioner, curtain, TV, light bulb and other equipment at home by making central equipment in the form of audio, TV box, etc. Specific sentences may be used in the test, thus directly omitting the wake-up command. However, because there is no wake-up command, it may be easier to trigger by mistake in actual use. However, issues such as false touch prevention should be considered later in mass production, but from the perspective of the testing machine, the experience of full voice control is quite cool and convenient.
Aim at the ecological chain of intelligent industry
In addition to providing relevant hardware and technical support, Cloud Zhisheng also hopes to form China's smart industry ecological chain, and announced the establishment of AI Club on the same day, intending to work with multiple smart home manufacturers to jointly build a smart home industry chain based on smart voice technology and AI, and realize the ultimate vision of resource sharing, mutual benefit and win-win results.
He said that he hoped to work with many partners to build a "Chinese version of Homekit" to jointly promote the landing and development of smart home products. At the same time, Yunzhisheng would also make its own contribution to building a "Chinese smart industry ecological chain".
As a general user, the biggest significance of the new remote recognition technology and AI Club is that the relevant technology can be applied to a variety of intelligent hardware more quickly, and with the hardware resources of Shenzhen, the hometown of intelligent hardware, this is likely to be a huge opportunity for China's intelligent hardware to catch up with and lead similar products abroad.
