Week 7: Music Power and Technology-Feminist Readings of AI Music 1

Vocabulary flashcards covering the intersection of music, power and technology, focusing on feminism, AI history, copyright eras, and digital neocolonialism.

Feminism

A way of analysing and critiquing systems of power and their impact on marginalised others, examining inequality and challenging what is treated as "normal" or "neutral."

AI generated music

Music created using datasets and algorithms, such as deep learning and neural networks (GANS, variational autoencoders, and transformers), to produce entirely new compositions or variations.

Era of the Copyists ($12 ext{th Century}$)

A period where church and monarchs controlled music to prevent the spread of heretical or undesirable material.

Era of the Printing Press ($15 ext{th Century}$)

A period that broadened music distribution but tightened regulation through royal privileges issued to select printers and publishers, creating monopolies.

Statue of Anne ($1710$)

A significant legal development that protected authors and enabled the expansion of the early music-printing industry.

French Copyright Law ($1793$)

Legislation that granted musicians the right to print, sell, and distribute their works.

Union of Authors, Composers and Music Publishers ($1853$)

An organization that enabled musicians to demand royalties for every performance, shifting power away from financiers and back toward creativity.

Era of the Record Company ($20 ext{th Century}$)

An era where record companies shaped the economy and accessibility of music through ownership of copyright and physical production.

Era of Tech Companies ($1980 ext{s}$ – $2000 ext{s}$)

A period where dominant technology companies mediated music consumption through CDs, cassettes, subscription-based streaming, and internet-connected devices.

Era of Generative AI ($2023$ - now)

A current era where commercial AI companies often circumvent copyright frameworks by training models on copyrighted music without obtaining prior permission.

Digital neocolonialism

The exploitation of labour, the environment, and data for the benefit of AI companies and their associated countries, mirroring European colonial resource extraction.

Environmental Impact of AI

Direct consequences including decimating forests, poisoning water sources, exploiting labour (children and climate refugees), and reactivating dormant nuclear plants.

MIR (Music Information Retrieval)

The academic field where AI music generation originates, focusing on dataset curation, annotation, and making music "computationally legible."

RWC (Real World Computing) Music Dataset ($2000$)

A dataset containing $328$ tracks (~$23$ hours of audio) across $5$ sub-collections, recorded specifically for research to avoid copyright issues.

GTZAN Dataset ($2002$)

A widely used MIR research dataset containing $10$ genres with $100$ clips each, noted for having $7.2$% duplicate recordings and $10.6$% mislabelled files.

Gender statistics in Music Production

Women comprise only $3.2$% of producers and hold only $5$% of production roles in commercial studios.

Gender statistics in AI Research

Women comprise only $12$% of contributions at major AI conferences and hold only $16.1$% of tenure-track faculty positions in AI.

Gender statistics in the AI Workforce

Women hold $22$% of AI roles globally, representing $15$% of AI research staff at Facebook and $10$% at Google.