Deep Bach

In this post we will try to introduce the Deep Bach! Hope you enjoy!

Introduction

DeepBach is an artificial intelligence that composes like Bach. DeepBach, a deep learning system based on recurrent neural network (RNN) architecture, has the unique capacity to produce music in the style of famed Baroque composer Johann Sebastian Bach. The system employs a graphical model to represent polyphonic music, with a particular emphasis on choral compositions. Graphical model developed to model polyphonic music, Using pseudo-Gibbs sampling and a customized representation of musical data, DeepBach does well.

The model is taught to recognize and mimic the intricate patterns and structures in Bach’s chorale harmonizations. DeepBach, once trained, can make extraordinarily convincing chorales in the style of Bach, a feat that sets it apart from other automatic music creation approaches, which typically function sequentially.

The level of control provided by DeepBach is one of its most notable characteristics. Users can impose restrictions on the created score, such as notes, rhythms, and cadences. This level of personalization enables a one-of-a-kind combination of human curiosity and AI efficiency.

In a blind test, roughly half of the participants thought a DeepBach-generated harmony was produced by Bach himself. This outcome is noticeably better than music generated by other algorithms.

Below is a sample of DeepBach’s work:

Remarks

Deep learning algorithms, such as DeepBach, have shown great potential in the realm of music generation. They take a fresh look at composition, fusing technology and art in an intriguing and novel way.

However, there are still obstacles to be conquered. One major difficulty is the creation of algorithms capable of producing really original music rather than simply copying existing compositions. Another problem is ensuring that the generated music is emotionally resonant and appealing to listeners as well as technically sound.

Despite these difficulties, the potential for this technology to transform the music industry is unquestionably fascinating. As the sector develops and improves, we may witness a new era of music production and consumption, fueled by AI’s unique talents.

Sources:

  • Sturm, B. L. (2016). “DeepBach: a Steerable Model for Bach chorales generation.” arXiv preprint arXiv:1612.01010.
  • Eck, D., & Schmidhuber, J. (2002). “Learning representations by back-propagating errors.” Cognitive computation.
  • https://www.sonycsl.co.jp/news/3882/