Speech recognition is evolving rapidly, can robots really understand us?

If a machine is to be able to understand human speech, it must go through a training process similar to that of a human. People first find the phonemes or word syllables that represent speech, along with their frequency, amplitude and time-varying characteristics, and store them in the machine. When the robot recognizes human speech, it first compares the parameters of the speech to be recognized with those inside the machine one by one, finds the most similar reference sound, and outputs the recognition result through display and sound. If there is no similar voice, it will refuse to recognize it. After all, a robot cannot compare itself to the "intelligence" of a human being. It is based on recognizing people's voices. However, people's voices vary from person to person, from urgent to slow, from high to low. The ambiguity of pronunciation is so great that it is difficult for robots to recognize it.

In recent years, the development of intelligent robots in China has shown rapid growth in momentum, which is represented by a small ⅰ-type robot launched by the intelligent customer service robot series. Using the self-developed robot platform, the small i robot for Dr. Fusion Hai Bao, Jiangsu Mobile, Tianjin Municipal Government, Lenovo Group, Bank of Communications and other 200 large-scale enterprises and the government launched a high-quality, high-efficiency intelligent customer service robot solutions, and has achieved a positive response in the community. Among them, the resolution rate of Dr. Fetion Haibao's problems is over 98%, and when chatting with Haibao, users feel the same as real people. Jiangsu Mobile's customer service robot was visited more than 200,000 times a day, and the robot handled 75% of the questions with an accuracy rate of more than 90%. Henan Mobile SMS robot handles more than 1 million SMS per day. Humorous and funny answers were once widely spread on microblogging and other online media.

Making robots understand human speech is a problem facing human technology. Because this voice technology has a wide range of applications, the use of this technology can automatically input human speech into the computer, automatically translated into the language of many countries, and exchange information with people of different countries and nationalities. In the field of automatic control, voice-activated squares and robots can be used to speak, which can reduce labor intensity and improve productivity. Researchers at home and abroad have made some achievements in this field.1223 An American company developed a voice-activated typewriter that could understand 40,000 isolated words. This was regarded as the arrival of the era of practical speech recognition system.In 1221, Chinese researchers built a speech recognition system that could recognize 4,333 Chinese sentences consisting of isolated syllables and had good knowledge learning ability.

Humans have tentatively mastered the technology that allows robots to understand human language, giving humans highly intelligent hearing abilities that adapt them to cold, emotionless machines. This technology will evolve rapidly, further realizing the higher concepts of scientists working on speech recognition technology!