In recent years, music shows have emerged in an endless stream. Whether it is a singer or a tutor, the goal is to find an evaluation method that can exclude other factors and return to music itself. Not long ago, CCTV’s singing program “Wanted to Live†invited such an objective and fair music review – the robot “Little Thirstyâ€.
This robot developed by the Institute of Automation of the Chinese Academy of Sciences will score every singer who walks into the "Golden Pavilion". Only a score of more than 80 points indicates that it is favored by the "Little Thirsty" teacher and has the opportunity to step out of the "Song Pavilion". "Meet the audience."
It’s no longer a new thing to score the singer. The scoring system in KTV can even list the singers who are similar to the singer’s voice. So what are the secrets of “Little Thirstâ€? "Traditional music scoring software generally simply compares the singer's pitch and standard sound for similarity. To put it bluntly, it is to see if you have a run, this single evaluation standard does not fully measure the quality of music. It is more difficult to understand the emotional information that the singer wants to convey.†Wang Jinqiao, a researcher at the Institute of Automation, Chinese Academy of Sciences explained that compared with the “small thirstâ€, it is the first time from pitch, range, tonality, rhythm. The six dimensions of language sense and music sense objectively judge the singing, taking into account the dimension of music appreciation and the difficulty of quantification. It is to evaluate art with technology.
"The art of robot evaluation is actually a very difficult thing. The biggest difficulty is how to let the robot automatically discover the elements of music in the United States." Wang Jinqiao told the "Economic Daily" reporter that we invited a large number of music experts to sing from multiple dimensions. Based on the scoring, on the basis of the hierarchical structure of the convolutional neural network, the music spectrum is hierarchically abstracted and concise, and it is automatically found that each sentence in the music contains the common beauty of the music. Thereby learning the rules of pitch, pitch, vibrato amplitude and vibrato frequency in music, and complete the aesthetic quantification of the whole piece of music.
At the same time, the layer-by-layer information abstraction feature of the neural network can help the robot to transfer from the scale to the singing technique to the emotional expression, and perform layer-by-layer quantization and analysis, so that the robot can relatively objectively learn the expert's evaluation of the music, and finally realize the pair of notes. "Understanding".
Because each singer differs in singing style and style, there will be a big difference between songs, but when the mass music expert score data is put together, it can basically represent the current mainstream music review experts' evaluation of music. . Based on these data, the music evaluation system based on artificial intelligence technology allows “Little Thirst†to judge the music objectively and scientifically.
In addition, the multi-dimensional speech features extracted by the convolutional neural network of the "small thirst" have precise quantitative characteristics, and the accuracy can reach 1‰ when the pitch information is quantized and distinguished. On this basis, combined with the expert's experience of the sound performance of the song sound as the supervision information, the neural network model can establish an accurate mathematical expression model for the pitch information, so as to quantify the pitch score.
The sound of the most sensational "music" also has a specific mathematical model support. As an art, music needs to convey certain emotions. Music is the sensory ability of people to understand this feeling. Expressiveness is the comprehensive performance of the singer's singing process. The current judging criteria are more dependent on the authoritative judgment of music experts. “Experts score the performance with 1 to 10 scores. We extract the characteristics of musical expressiveness through convolutional networks and two-way long-term and short-term memory models. This feature can fully analyze the expressive power of music in the time dimension. The simulation expert judges the expressiveness of music," said Wang Jinqiao.
Judging from the multi-period programs that have been aired, "Little Thirsty" can give a fair evaluation of various music singing methods, such as Qin Qiang, rap, and English songs. At the scene, the "Little Thirsty" is directly connected to the singer's microphone, and any small flaw can be captured by it. "In the specific scoring process, the robot is listening to a sentence, the expert may not even hear it, but the robot does not exist, and will not be affected by the atmosphere and the environment." Wang Jinqiao said, "small "Thirsty" is the same as the singer's equal treatment, and the fairness and reasonableness of the scores are also recognized by the audience. Everyone affectionately calls it "the most calm review in history."
"'Little Thirst' is still a junior version. We will introduce more algorithms and more dimensions to evaluate music. We will also try to make it learn to create music and make a wonderful combination of touching notes." Talking about the future of "Little Thirsty", Wang Jinqiao and his team hope to promote it to various music scoring systems, such as singing, KTV and so on. Singing enthusiasts can combine their "small thirst" evaluation to continuously improve their singing. One day, maybe the "Little Thirst" will be upgraded from "Music Review" to "Dream Mentor". (China Economic Net)
Our company offers rubber, and specialty seal and o-ring for molding electronic and electrical Connectors, both circular and rectangular types. We have in house capabilities and quick turn around.
Cable Grommets with various Grommet Shapes, Rib Styles, Slot Shapes, Inner Flanges, and Colors - To have access to our 3D Cable Grommet Design Tool
Silicone Rubber Products,Cable Silicone O-Ring,Rubber Seal,Custom Silicone Seal,Waterproofing O-Ring,Tpe Grommet
ETOP WIREHARNESS LIMITED , https://www.etopwireharness.com