Melody MIDI accompaniment generation method based on deep neural network

Abstract

The invention discloses a melody MIDI accompaniment generation method based on a deep neural network, and belongs to the technical field of artificial intelligence and music science and technology. The melody MIDI accompaniment generation method comprises the steps of collecting an MIDI data set with a body tailoring label, processing the MIDI data set to obtain a data set, encoding through MuMIDIto obtain entries, inputting the entries into a GC-Transformer model for training until a total loss function converges, completing training of the GC-Transformer model, finally encoding MIDI fragments only containing melody, and inputting the MIDI fragment into a trained GC-Transformer model, and finally outputting the generated accompaniment MIDI fragment. The melody MIDI accompaniment generation method has the advantages of short generation time, high generation quality, less hardware resource consumption, less training data volume and the like.

Description

technical field [0001] The invention relates to the technical fields of artificial intelligence and music technology, in particular to a method for generating melody MIDI accompaniment based on a deep neural network. Background technique [0002] Art creation has always been considered the exclusive domain of artists. However, in recent years, with the development of deep learning, art creation has made great progress and reached unprecedented heights. For example, it can generate paintings of a specified style, or generate A musical composition that would pass the Turing test. Music generation is a huge field, and it faces many challenges, especially when users want to control many attributes of music generation (such as the number of generated instruments, music genres, etc.), the resulting music has a lot of room for improvement, This kind of task is collectively referred to as conditional controllable music generation. [0003] The current mainstream condition-controll...

AI Technical Summary

Problems solved by technology

This work can achieve multi-track accompaniment generation, but its disadvantage is that it cannot generate music with controllable length and genre

The third type of work combines the functions of the above work, and can realize multi-track music with controllable length and genre, a typical example is MuseNet; but it requires a large amount of MIDI as a training set (order of magnitude is millions), and Due to the large amount of training parameters, the time cost of generating a specified song will be very high

At the same time, this work has no way to realize the functions of segment continuation and melody accompaniment. It can only generate music segments at a specified time from the beginning, and in the demo shown, the note density of each track is relatively sparse, and the quality of the music is low.

To sum up, if you need to make a multi-track music clip with a style controllable and any length in a short period of time, there is no ready-made solution on the market that can meet all the above requirements

Images