The Computer Music Gender Gap

Analysis of author demographics in major Computer Music conferences revealing female representation consistently below 20% (but some signs of minor improvements).

Figure 1: A flowchart depicting the project workflow, from retrieval of proceedings data to gender prediction based on name.

Figure 1: Project workflow.

Figure 2: Barplots showing percentage of female author names for all conferences until year 2016.

Figure 2: Percentage of female author names in th three major Computer Music conferences (until 2016).

Figure 3: Barplot showing the percentage of female versus unknown author names until year 2021, for the ICMC conference proceedings.

Figure 3: Female vs. unknown author names in ICMC proceedings (until 2021).

Video 1: Sonic representation of the gender gap in the ICMC dataset until 2021. Head to 54.10 to hear what it would sound like if the gender (im)balance would be reversed.

Video 2: Sonic representation of gender gap in the ICMC dataset until 2016 (simple sound model).

Video 3: Video 2: Sonic representation of the gender gap in the ICMC dataset until 2016 (granular sound model).

Keywords

Music Tech, Data Science .

Links

Publication (2021), publication (2017), sound examples (2021), sound examples (2017).

Technologies

Python (numpy, pandas, genderize), LaTeX and BibTex, R (ggplot2, plotly), SuperCollider (sound synthesis).

Background

Despite early contributions from pioneers like Ada Lovelace and Grace Hopper, gender imbalance remains a persistent challenge in STEM fields, including in Computer Music. This project aims to quantify and analyze the extent of this gender gap across key Computer Music conferences using data-driven methods.

Aim

The goal of this project was to quantify the representation of female authors in the proceedings of three key conferences in the Computer Music field: Sound and Music Computing Conference (SMC), New Interfaces for Musical Expression (NIME) conference, and International Computer Music Conference (ICMC).

Approach

In the absence of a readily available dataset, I developed one by extracting information from external sources, including PDF proceedings, conference webpages, and BibTeX files, utilizing custom Python scripts for data scraping. To engineer the 'author gender' feature, I used the genderize.io API for gender classification, categorizing names as 'male,' 'female,' or 'unknown.' I then conducted additional manual research to refine the data for 'unknown' classifications.

To analyze the gender representation trend over time, I applied an Autoregressive Integrated Moving Average (ARIMA) model.

To further illustrate the skewed distribution of author genders— the majority was still predominantly male—I transformed the data into musical sounds through a technique known as sonification.

Findings

Analysis of the data revealed that the ratio of female authors remained consistently below 20% until 2016. This trend largely persisted up to 2021, indicating a long-standing gender imbalance in the field. However, a closer examination of the International Computer Music Conference (ICMC) proceedings showed signs of gradual improvement, with female author representation increasing from 2.8% in 1981 to 16.2% in 2017. Notably, this conference exhibited a statistically significant annual increase of 0.17% (± 0.1) in female author representation (p < .001), suggesting a small positive trend.

The findings were documented in two publications and presented at several conferences and seminar series, generating engaging discussions within the research community. These dialogues have contributed to ongoing efforts to promote greater inclusivity in the Computer Music field.

Return to main page.

Page updated

Google Sites

Report abuse