Although artificial intelligence may replace people's work, it can also be a tool for inspiring artistic creativity. Google recently announced the NSynth Super sound synthesizer, which uses the sound of a real instrument to create more new sounds through AI. Creation provides more fun and possible. According to CNBC report, NSynth Super is a new achievement of Magenta's R&D program under Google Brain. In April 2017, researchers first published the NSynth algorithm, which uses a neural network (neural network). ) Train more than 100,000 sounds from different instruments, and generate new sounds based on these sound characteristics. Now the NSynth Super is based on the hardware prototype created by this technology. Using the open source software version of NSynth You can select two kinds of instruments first, then adjust the proportions of the sounds of these two kinds of instruments. Then the system will create new sounds. Users can use the keys on the computer keyboard to play the notes. New sounds are played. As for the hardware version of the NSynth Super prototype, it provides the operation of the physical knobs and touch screens. The number of choices is increased to four. After selecting the instrument by using the knobs in the four corners of the screen, and then using the finger to slide in the four quadrants on the touch screen to adjust the ratio of the sounds of various instruments, you can synthesize more. The new voice. In addition, NSynth Super also has the same functions as general synthesizers that can adjust basic parameters such as attack, decay, sustain and release (ADSR). Joao Wilber, a researcher at Google London Creative Labs who participated in this hardware development. He pointed out that this project hopes to provide tools for music in the music field that are convenient for people to sway creativity, but Douglas Eck, a senior researcher of the Magenta team, said that NSynth Super has an elegant metal shell appearance and is internal to Google. The manufacturer assists with the production, but Google itself does not intend to market the hardware in mass production. In the development process of NSynth, Google Brain also cooperated with DeepMind, an AI research company owned by the company's parent company Alphabet, and the researchers have related hardware design and open source software. Released on GitHub, players can use a simple Raspberry Pi mini computer with a variety of materials or 3D printing to create Its own exclusive NSynth Super sound synthesizer. Magenta is an open source project launched by Google Brain in June 2016. It is dedicated to using AI technology to provide artists and musicians with better creative tools and more inspiration. Magenta's other research is also It includes AI Duet, which can create melody based on the melody played by the player, and Sketch-RNN that can imitate human graffiti.
