Xiaomi's AI strength has been verified again, and its voice recognition algorithm capability ranks among the best in the world.

Recently, Xiaomi has made the latest progress in voice recognition algorithms. Its algorithm model is ahead of other institutions in terms of performance and ranks at the top of international rankings. This breakthrough was achieved by Xiaomi in the AudioSet audio labeling task released by Google

Google divided the AudioSet data set into three subsets. The first two subsets were used for training and were merged and called "AudioSet-2M". It was in this merged training set that Xiaomi's voice recognition algorithm model broke through 50 mAP in the industry, refreshing audio marking technical indicators. In addition, Xiaomi also released a Mini version of the model, suitable for resource-constrained scenarios.

The breakthrough of this technology means another breakthrough in Xiaomi's sound recognition algorithm capabilities. After Xiaomi's many smart hardware devices apply this technology, they can capture and identify environmental sounds more keenly, greatly improving the intelligence of the hardware. This creates a more convenient smart life experience for users.

At the same time, Xiaomi’s audio tagging technology capabilities under the voice recognition algorithm have effectively promoted the research and development process of cutting-edge research in the industry.

Xiaomi’s improved voice recognition algorithm can be widely used in Xiaomi’s smart devices. The audio tagging algorithm can identify a wide range of environmental sounds, such as babies crying, animal sounds, car engines, explosions, smoke alarms, doorbells, water flows, etc., and allow the sounds in the environment to be expressed in modalities such as text, Let the sound be "seen". This can make hardware devices smarter and bring users a more efficient and accurate voice recognition experience.

The "Xiaomi Hearing Sound" function in Xiaomi mobile phones can monitor 14 important environmental sounds including fire alarms, babies crying, kettle boiling sounds, etc., and push text messages through the mobile phone notification bar. This function effectively helps the lives of hearing-impaired people, and the Xiaomi bracelet can also display and receive these sound information simultaneously

This algorithm is more widely used in smart home devices. Currently, Xiaomi's voice recognition algorithm has been used in many smart home devices, such as the baby cry monitoring function of Mijia cameras, which can push notifications to users' mobile phones in real time when it detects a baby crying. The Xiaomi Sound speaker is also equipped with a sound recognition function, which can identify six types of home environment sounds that users care about, including home alarms, crying babies, fire alarms, running water, cats meowing, and dogs barking. Xiaomi Health APP’s sleep snoring monitoring can help track users’ snoring and sleep talking during sleep.

In order to better realize the potential of this algorithm, Xiaomi provides a broad space for the application of this algorithm technology through its AIoT platform ecosystem. At present, Xiaomi's acoustic voice team has fully applied its self-developed acoustic voice technology to more than 5,000 smart products in 79 categories including Xiaomi mobile phones, speakers, TVs, headphones, watches, and robots.

In the field of artificial intelligence, which belongs to the voice recognition algorithm, Xiaomi is actively involved in it, as one of the bottom tracks that Xiaomi has continued to invest in for a long time. It is reported that since Xiaomi established its AI team in 2016, the AI ​​team has expanded six times in seven years and has reached more than 3,000 people. At the same time, Xiaomi's AI technical capabilities have also covered many directions such as vision, acoustics, speech, NLP, knowledge graphs, machine learning, large models, and multi-modality, and have fully empowered many business sectors such as mobile phones, automobiles, AIoT, and robots. . In April this year, Xiaomi officially established an AI large model team to fully embrace large models, and there are currently some application attempts. The first application model is to upgrade the intelligent voice assistant Xiao Ai to a large model version and open invitation testing.

The above is the detailed content of Xiaomi's AI strength has been verified again, and its voice recognition algorithm capability ranks among the best in the world.. For more information, please follow other related articles on the PHP Chinese website!