Composing Heavy Metal with Artificial Intelligence - GPT-2 Trained on Music - Demo

Tristan Behrens
21 Jan 202215:06

TLDRTristan demonstrates how he composes music with AI, showcasing his process of creating heavy metal music using a GPT-2 model trained on thousands of songs. In 2021, he released three albums with AI-generated notes, and in 2022, he developed a web-based tool for composing, enhancing his workflow. The video illustrates the step-by-step creation of a heavy metal track, from drums to guitars, bass, and even piano, all generated and curated by AI, highlighting the efficiency and creativity AI brings to music composition.

Takeaways

  • 🎼 The presenter, Tristan, has used artificial intelligence to compose heavy metal music, releasing three albums in 2021.
  • 🤖 Tristan incorporated AI into his composing pipeline, allowing the AI to generate notes for 36 songs across three albums.
  • 🎶 The AI, based on GPT-2, was trained on 7000 heavy metal songs, enabling it to predict harmonious musical bars that fit with existing composition.
  • 🔧 In 2022, Tristan developed a new web-based tool using HTML5, JavaScript, and CSS to replace Jupyter notebooks for music composition.
  • 🚀 The new web tool offers enhanced functionality and a more user-friendly interface for composing music with AI.
  • 🥁 The demonstration starts with composing drums, suggesting the importance of rhythm in heavy metal music composition.
  • 🎸 Tristan then adds guitar, bass, and even piano parts to the composition, showing the AI's versatility in different instruments.
  • 🔄 The AI can adjust note density, affecting the complexity and rhythm of the music, allowing for dynamic composition.
  • 👩‍💻 The process involves curating AI-generated ideas, keeping what works and replacing what doesn't, providing a collaborative approach to music creation.
  • 📈 Tristan's use of AI has significantly reduced the time required to create a song structure, from one working day to half that time.
  • 🌐 The video invites viewers to engage by suggesting genres for future compositions and encouraging feedback in the comments section.

Q & A

  • What is the main topic of the video?

    -The main topic of the video is composing heavy metal music with artificial intelligence, specifically using a GPT-2 model trained on heavy metal music.

  • How many albums did the speaker release in 2021 using AI for music composition?

    -The speaker released three albums in 2021 using AI for music composition.

  • What was the speaker's previous method for composing music with AI?

    -The speaker previously used a tool based on Jupyter notebooks with widgets to compose music with AI.

  • What new development did the speaker work on for composing music in 2022?

    -In 2022, the speaker developed a new web-based composer using HTML5, JavaScript, and CSS, which replaced the Jupyter notebook method.

  • What is the role of the deep neural network in the music composition process shown in the video?

    -The deep neural network, which is GPT-2 trained on heavy metal songs, predicts and generates bars of music that fit harmoniously with the existing composition.

  • How does the speaker use the AI-generated music in their workflow?

    -The speaker uses the AI-generated music as a starting point, curating the ideas by keeping what they like and replacing what they don't with other ideas generated by the AI.

  • What is the significance of 'note density' in the context of the video?

    -Note density refers to the number of notes per bar in the music composition, affecting the complexity and rhythm of the piece.

  • How does the speaker incorporate different instruments into the composition?

    -The speaker starts with a basic rhythm, such as drums, and then adds other instruments like guitar and bass, adjusting and recomposing as needed to create a cohesive piece.

  • What is the advantage of using AI for music composition as demonstrated in the video?

    -The advantage is that AI can generate musical ideas quickly, allowing the composer to focus on curating and refining the AI's output rather than manually composing every note.

  • What is the speaker's process for turning the AI-generated music into a complete song?

    -The speaker takes the AI-generated music, rearranges it, adds details, and structures it into a complete song with elements like verses, choruses, bridges, intro, and outro.

  • How does the speaker engage with the audience for future music composition sessions?

    -The speaker encourages the audience to comment with their preferences for heavy metal music genres, and may create future sessions based on audience feedback to compose different styles of metal music.

Outlines

00:00

🎼 AI in Music Composition

The speaker, Tristan, discusses his experience using artificial intelligence to compose music, specifically in the heavy metal genre. He details his process of integrating AI into his music composition pipeline, using a deep learning network trained on heavy metal music. In 2021, he released three albums totaling 36 songs with AI assistance. He also shares his transition from using Jupyter notebooks to developing a web-based composer using HTML5, JavaScript, and CSS, which includes new functionalities. The speaker demonstrates the initial steps of composing a new project called 'Friday,' starting with drum sequences and adjusting note density to create a heavy metal rhythm.

05:01

🥁 Developing a Heavy Metal Composition

Tristan continues his discussion on music composition by adding guitar and bass to the drum sequence he created earlier. He uses the AI tool to generate musical ideas, which he then curates and refines. The speaker emphasizes the ability to visually inspect the music notation and listen to the AI-generated segments, allowing for an intuitive understanding of the composition. He adds a second guitar, adjusting the note density to create a different texture, and then introduces a piano accompaniment to enrich the music's texture further. The speaker showcases the flexibility of the AI tool in composing various instruments and adjusting the music's density and intensity.

10:01

🎹 Curating AI-Generated Music

The speaker illustrates the process of curating AI-generated music, using the example of adding and adjusting drums, guitar, and bass. Tristan emphasizes the AI's ability to consider the context of existing musical elements when generating new bars of music. He demonstrates how to refine the composition by adjusting the intensity and note density, as well as experimenting with different instruments like piano. The speaker also discusses the efficiency of using AI for music composition, stating that it has reduced the time required to develop a song structure. He invites viewers to share their preferences for heavy metal sub-genres for potential future compositions.

15:03

📢 Conclusion and Call to Action

In the final paragraph, Tristan concludes his demonstration of using AI for music composition and invites viewer engagement. He encourages viewers to like the video, subscribe to his channel, and share their ideas in the comments section. The speaker highlights the user interface of the AI tool trained on heavy metal music and expresses hope that the audience appreciates his presentation.

Mindmap

Keywords

💡Artificial Intelligence

Artificial Intelligence (AI) refers to the simulation of human intelligence in machines that are programmed to think like humans and mimic their actions. In the context of the video, AI is used to compose music, specifically heavy metal, by generating notes and melodies. The script mentions the use of a deep neural network trained on heavy metal music to assist in the creative process, demonstrating how AI can be integrated into artistic endeavors.

💡Composing

Composing is the process of creating a piece of music by putting together various musical elements such as melody, harmony, and rhythm. The video script discusses the composer's process of creating music using AI, where the AI generates musical sequences that the composer then refines and incorporates into a full song. This showcases the collaborative potential between human creativity and AI technology in the field of music composition.

💡Deep Neural Network

A Deep Neural Network (DNN) is a type of artificial neural network with multiple layers between the input and output layers, allowing it to model and understand complex patterns. In the video, the DNN is trained on heavy metal music and is used to generate musical notes and sequences. The script illustrates the network's role in the creative process by predicting and composing bars of music that fit with the existing composition.

💡GPT-2

GPT-2 is a language model developed by OpenAI that is capable of generating human-like text. In the video, GPT-2 is mentioned as the underlying technology used to train the AI on heavy metal music. The script explains that the AI, based on GPT-2, is invoked to predict and compose music bars, taking into account the musical context to ensure harmony and coherence within the composition.

💡Heavy Metal Music

Heavy Metal Music is a genre of rock music characterized by aggressive vocals, heavy guitar riffs, and powerful drumming. The video's theme revolves around composing this genre of music using AI. The script describes the process of training the AI on heavy metal songs and using it to generate music that fits within this style, highlighting the AI's ability to understand and replicate genre-specific musical elements.

💡Note Density

Note Density refers to the number of musical notes per unit of time in a piece of music, often measured per bar. In the script, the composer adjusts the note density to control the complexity and intensity of the music being generated by the AI. By increasing or decreasing note density, the composer can influence the rhythmic feel and the overall texture of the composition.

💡Jupyter Notebooks

Jupyter Notebooks is an open-source web application that allows users to create and share documents containing live code, equations, visualizations, and narrative text. In the video, the composer mentions previously using Jupyter Notebooks with widgets for music composition, but has since transitioned to a web-based interface due to the limitations and personal requirements for more functionality.

💡Web Technology

Web Technology encompasses the different tools and platforms used to create and manage websites, including HTML5, JavaScript, and CSS. The script describes the composer's development of a new music composition tool using web technology, which offers a more dynamic and interactive interface for composing music with AI compared to the previously used Jupyter Notebooks.

💡Drum Sequence

A Drum Sequence is a series of drum beats arranged in a specific order to create a rhythmic pattern. In the video, the composer starts the composition process by generating a drum sequence with the AI, which serves as the foundation for the rest of the music. The script illustrates the use of AI to create a drum sequence that inspires the development of the rest of the composition.

💡Music Interface

A Music Interface is a user interface designed for creating, editing, or manipulating music. In the context of the video, the composer has developed a new music interface using web technology that allows for more intuitive and visually engaging music composition with AI. The interface enables the composer to easily add, modify, and delete musical elements such as drums, guitar, bass, and piano.

💡Piano Accompaniment

Piano Accompaniment refers to the part of a musical composition played by a piano, typically providing harmonic support and rhythmic structure to the main melody or other instruments. In the script, the composer uses the AI to add piano accompaniment to the heavy metal composition, demonstrating the AI's versatility in generating different instrumental parts that complement the overall piece.

Highlights

Tristan demonstrates the use of AI in composing heavy metal music with GPT-2.

In 2021, Tristan released three albums composed with AI, totaling 36 songs.

The AI was integrated into Tristan's composing pipeline, providing notes for song generation.

The AI, trained on heavy metal music, is used for creating the next album in 2022.

Tristan discusses tool development and transitioning from Jupyter notebooks to web technology.

A new web-based composer is introduced, featuring HTML5, JavaScript, and CSS.

The composer allows for the creation of new music projects with various settings like genre and note density.

Starting the composition with drums is suggested as a common practice in heavy metal.

The AI-generated drum sequence is played, inspiring further composition.

Tristan manipulates the drum sequence to create a rhythmic foundation for the song.

Guitar is added to the composition, with the AI providing riffs and chords.

The AI's ability to consider existing musical elements when generating new parts is highlighted.

A second guitar part is added, demonstrating the AI's versatility in note density.

The addition of a piano accompaniment shows the AI's capability to compose beyond traditional heavy metal instruments.

Tristan discusses the iterative process of composing, where the AI generates ideas that are curated by the composer.

The video concludes with Tristan reflecting on the efficiency gains from using AI in his music composition process.

Viewer engagement is encouraged through likes, subscriptions, and comments for genre preferences.