Voice intelligent AI TV will become the mainstream of the industry

2017 is the first year of AI color TV products-the so-called AI color TV is a new TV product that can understand everyone's language, execute language commands, and even freely talk to everyone. Statistics show that nearly a quarter of the newly launched color motor models in 2017 are equipped with voice intelligence.

In early 2018, the voice intelligent AI TV was further upgraded. The industry predicts that the proportion of new color TV products equipped with AI voice technology in 2018 will rapidly rise to more than 80%, that is, except for a few products that focus on low prices, most new products will become smart TVs. Intelligent voice is entering a "new stage" of rapid development.

Voice intelligent AI TV will become the mainstream of the industry

Consumer electronics, voice technology becomes standard

On January 29, as the second new product launched by Xiaomi TV in 2018, Xiaomi Box 4 was officially released. Its biggest feature is the direct preset artificial intelligence voice system. Smart box products are a key tool for "existing old TV experience upgrade". About half of domestic color TV users use smart boxes, OTT boxes, IPTV boxes and other products to access TV networks. Xiaomi Box 4 marks that this "stock" market has entered a key new stage of "smart voice" development.

Another new consumer product of intelligent voice technology is the "smart speaker". At CES 2018, Baidu displayed 3 smart speaker products. The product is also an indisputable place for international giants such as Samsung, Apple, and Google. In the domestic market, the Tmall series also launched smart speaker products in mid-2017. At present, the best development of the smart speaker industry is the US market: in the United States, nearly one-sixth of people (more than 50 million) owned smart speakers in 2017, which has more than doubled from 16 years.

From the perspective of the domestic market, smart voice boxes, smart voice speakers and smart voice TVs have constituted the three major platforms for home and personal consumption of "smart voice" applications. And the potential market size of each category is "10 million units". However, the sales of these three categories and the king of smart voice applications: smartphones are really "small". Mobile phone products have been using voice input, voice operation and other technologies for a long time. Among them, more than 60% of users often use voice input. However, as a handheld device, the application of voice operations on mobile phones is not obvious.

In addition to the above consumer electronics devices, the home appliance industry is also optimistic about the development of "home appliance embedded intelligence" technology such as the voice intelligence of "air conditioning", the sound perception of range hoods, and the automatic operation of fume sensing. A person with air-conditioning said that the combination of voice-controlled air-conditioning products and automatic environmental sensing technology can bring a better operating experience, a better no-operation automatic environmental control experience, and the energy-saving effect of avoiding idling.

As the most important communication method for human beings, the development of voice technology has obviously become an important innovation ring in the consumer electronics industry. The colorization of color TV products has become an important industry trend, and is rapidly becoming popular.

Voice intelligence is amazing, how to make it really easy to use

Many people think that smart TV voice technology is "installing more software". However, real product development is much more complicated than "one software".

To achieve voice interaction, you first need to equip the device with a "microphone", that is, a "pickup" product. Unlike our general conference speech and KTV entertainment microphones, the microphones of smart voice TVs must have three technical characteristics. The first is that the far-field effect is good. That is, at least 5 meters away, it can accurately identify daily conversations, and even identify dialogues with low voice but clear. This is very different from the application of mobile phones, KTV, live broadcast, and conference room microphones "to the mouth".

Second, the smart voice technology microphone must be "omnidirectional". Since voice communication is not "mouth blowing into the microphone", the "angle relationship" between the communication personnel and the microphone and other devices is very different. For example, a smart speaker placed on the coffee table, then he needs to be able to recognize the sound within 360 degrees. Therefore, intelligent voice products usually use a "multi-microphone coupling" array to achieve omnidirectional sound pickup.

Third, the microphone of the intelligent voice technology must have a high “signal-to-noise ratio”. That is to be able to resist the interference of multiple noises. Far-field and omnidirectional pickups further increase the test of the microphone's signal-to-noise ratio.

However, for the intelligent voice equipment, its biggest technical requirements are obviously not in "harsh sound pickup technology", but in "speech to semantic" understanding. The purpose of the intelligent voice device is to understand the specific meaning of the other party. This is completely different from the traditional conference, KTV, telephone and other scenes, where the microphone and voice system are only responsible for "moving the sound as it is".

What are the difficulties in recognizing the meaning of speech? For example, it is necessary to effectively distinguish "who is in conversation with who is in conversation" from multi-person voices, especially in multi-person conversations in a home environment, and the sound of the audiovisual content being played. This involves an identification and authentication technology called voiceprint (similar to the functions of fingerprint recognition, face recognition, etc.). As another example, not everyone is a "newscast" standard mandarin user: accents, dialects, mantras, these treatments are very "personal" and "differential". This requires the intelligent voice system to be able to learn and upgrade itself and adapt to the language characteristics of the "master". The latter involves advanced intelligent technologies such as machine learning.

Of course, the TV set with intelligent voice technology is not "a person is fighting", this kind of TV set is a kind of "network TV set" at first. That is, voice technology can get more powerful "support" from the Internet. Including, the upgrade of smart programs and algorithms, remote access and use of personal language features and habits, remote cloud computing provides more powerful algorithms and support for understanding computing power, and supports the development of more "potential applications".

In summary, the upgrade of AI voice TV is not a simple addition of software. With the further development of voice and artificial intelligence technology, it is the general trend to introduce special intelligent coprocessors, neuron processors, and change the underlying architecture of traditional products. The integration of voice technology and color TV products will be a systematic product technology system innovation.

With AI voice, what does color energy do?

The enthusiasm of color TV and other consumer electronics and home appliance companies in AI voice technology is enough to explain the "strategic" and "important significance" of this technological progress.

For now, the AI ​​voice function of color TV is more of a substitute for the "remote control". Many times it is "very annoying" to find the remote control, the remote control is broken, and the remote control is out of power. The color TV industry has been trying to find a "remote control replacement". In the early days, body language and air touch technology were developed. However, this kind of technical experience accuracy and extensive interaction are not friendly and have not become mainstream.

Products using voice technology can not only directly call the content of color TV menus, program menus in TV stations and TV networks, but also enable text input and direct query of "non-standard menu" content indexed by text information on the network. In other words, voice technology not only "replaces all the functions of the remote control", but also "creates functions that the remote control is not good at (such as text input) and non-existent (voice dialogue)."

From the perspective of future development, voice intelligence technology will open up some new application scenarios for color TV products. The role of traditional color TV is mainly video entertainment-that is, the content is always transmitted in one direction. However, voice technology can make color TV a "key node." As a voice interaction and information processor, color TV can become the "central control brain" of other smart home products. As an input terminal connected to the cloud system, the color TV can rely on the more powerful computing and knowledge energy behind it, and "talk to the audience, pull homely", and even achieve more active behavioral functions, a large number of Internet applications and interfaces ...

In a sense, the color TV with intelligent voice technology is just a window into the era of "human-machine intelligence". Taking this as a clue, it involves the entire smart network of home, life, family and personal "smart applications"; and the cloud storage, computing and knowledge system that this network depends on. Color TV companies highly value the development of AI TV precisely because of the huge “potential possibilities” behind voice TV.

Even, the innovation of speech and semantic technology has become an important national strategy. In December 2017, the Ministry of Industry and Information Technology issued the notice of the "Three-year Action Plan for Promoting the Development of the New Generation Artificial Intelligence Industry (2018-2020)". Among them, it is mentioned that "support the innovative application of new generation speech recognition framework, colloquial speech recognition, personalized speech recognition, intelligent dialogue, audio and video fusion, speech synthesis and other technologies." "In 2020, the average accuracy rate of Chinese speech recognition in multiple scenarios will reach 96%, the recognition rate of the 5-meter far-field recognition will exceed 92%, and the accuracy rate of user dialogue intention recognition will exceed 90%"; The translation accuracy rate of the products in the English-to-English and English-to-Chinese translation scenarios exceeds 85% ";" Important goals and requirements such as the penetration rate of the smart TV market reaching more than 90% ".

The smart TV revolution that began in 2010 has made TV surfing a popular and main way of experience; now the smart revolution has entered the second phase of "human-machine intelligence" and is using voice interaction as the initial key to open the door to the future of the color TV industry.

Powell Prism

Powell Optical glass prism is also called laser linear generator lens. Unlike cylinder lens, the dome of Powell optical glass prism is actually a complex two-dimensional aspheric curve, which will produce a lot of spherical aberration and redistribute light along the line; Reduce the light in the center area while increasing the light level at the end of the line, transforming the laser beam into a uniform straight line. Cylindrical Lens will produce poorly illuminated lines with fading edges and hot spots in the center, while the straight lines produced by Powell Optical glass prisms have uniform energy distribution and uniform light distribution along the line.The Optical Prisms provide uniform energy distribution along the entire length of the line.


Powell Optical glass beamsplitter are currently widely used in industrial and research applications, where uniform intensity distribution is essential, because lasers are used as illumination sources, or, for example, in material processing, when the entire area irradiated by the laser must be processed to be uniform and equal The way. Powell lenses will produce a very uniform line of illumination for a variety of machine vision applications; from biomedical and automotive assembly to chocolate chip cookie production.

Optical Powell Lens,High Precision Powell Lens,Powell Prism,Bk7 Powell Prism

Zoolied Inc. , https://www.zoolied.com