By the same authors

Algorithmic music curation in audio branding: Commercial and critical perspectives on an international research and innovation project

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Standard

Algorithmic music curation in audio branding: Commercial and critical perspectives on an international research and innovation project. / Edwards, James; Egermann, Hauke Wolfgang.

Algorithmic Music: Value, Creativity and Artificial Intelligence A One-day Symposium. 2019.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Harvard

Edwards, J & Egermann, HW 2019, Algorithmic music curation in audio branding: Commercial and critical perspectives on an international research and innovation project. in Algorithmic Music: Value, Creativity and Artificial Intelligence A One-day Symposium.

APA

Edwards, J., & Egermann, H. W. (2019). Algorithmic music curation in audio branding: Commercial and critical perspectives on an international research and innovation project. In Algorithmic Music: Value, Creativity and Artificial Intelligence A One-day Symposium

Vancouver

Edwards J, Egermann HW. Algorithmic music curation in audio branding: Commercial and critical perspectives on an international research and innovation project. In Algorithmic Music: Value, Creativity and Artificial Intelligence A One-day Symposium. 2019

Author

Edwards, James ; Egermann, Hauke Wolfgang. / Algorithmic music curation in audio branding: Commercial and critical perspectives on an international research and innovation project. Algorithmic Music: Value, Creativity and Artificial Intelligence A One-day Symposium. 2019.

Bibtex - Download

@inproceedings{8e5b608c71a04d33a4b3132937d6a758,
title = "Algorithmic music curation in audio branding: Commercial and critical perspectives on an international research and innovation project",
abstract = "Audio branding is a marketing technique based on the insight that music forges connections between people and symbols, including brands. The European Commission-funded research project ABC_DJ has developed an algorithmic music recommender system specifically for use in audio branding scenarios. The ABC_DJ team first developed a validated list of marketing-relevant musical attributes comprising 36 terms (such as innovative, joyful, and authentic). Two international listening experiments followed, with a total of n=10,144 participants rating over 500 excerpts of music using these terms. The same pieces were then analysed on an acoustic and musical level using music information retrieval tools. Finally, machine learning was employed to map the correlations between acoustic features and ratings of semantic attributes. ABC_DJ furthermore addressed social variance in musical experience by incorporating values-based segmentation into the second listening experiment. The result is an algorithm capable of predicting perceived musical expression and liking based on the acoustic features of music pieces on the one hand and the characteristics and values of listeners on the other. In a commercial context, the resulting ABC_DJ system will allow brands to search digital music libraries in real-time, automatically updating their track pools at a pace commensurate with that of contemporary cultural production. The proposed presentation will demonstrate the ABC_DJ system. It will also introduce critical perspectives derived from qualitative interviews with n=23 expert users, who voiced attitudes ranging from scepticism in the aesthetic quality of algorithmically-curated playlists to fear that systems like ABC_DJ might devalue the labour of human music curators.",
author = "James Edwards and Egermann, {Hauke Wolfgang}",
year = "2019",
language = "English",
booktitle = "Algorithmic Music: Value, Creativity and Artificial Intelligence A One-day Symposium",

}

RIS (suitable for import to EndNote) - Download

TY - GEN

T1 - Algorithmic music curation in audio branding: Commercial and critical perspectives on an international research and innovation project

AU - Edwards, James

AU - Egermann, Hauke Wolfgang

PY - 2019

Y1 - 2019

N2 - Audio branding is a marketing technique based on the insight that music forges connections between people and symbols, including brands. The European Commission-funded research project ABC_DJ has developed an algorithmic music recommender system specifically for use in audio branding scenarios. The ABC_DJ team first developed a validated list of marketing-relevant musical attributes comprising 36 terms (such as innovative, joyful, and authentic). Two international listening experiments followed, with a total of n=10,144 participants rating over 500 excerpts of music using these terms. The same pieces were then analysed on an acoustic and musical level using music information retrieval tools. Finally, machine learning was employed to map the correlations between acoustic features and ratings of semantic attributes. ABC_DJ furthermore addressed social variance in musical experience by incorporating values-based segmentation into the second listening experiment. The result is an algorithm capable of predicting perceived musical expression and liking based on the acoustic features of music pieces on the one hand and the characteristics and values of listeners on the other. In a commercial context, the resulting ABC_DJ system will allow brands to search digital music libraries in real-time, automatically updating their track pools at a pace commensurate with that of contemporary cultural production. The proposed presentation will demonstrate the ABC_DJ system. It will also introduce critical perspectives derived from qualitative interviews with n=23 expert users, who voiced attitudes ranging from scepticism in the aesthetic quality of algorithmically-curated playlists to fear that systems like ABC_DJ might devalue the labour of human music curators.

AB - Audio branding is a marketing technique based on the insight that music forges connections between people and symbols, including brands. The European Commission-funded research project ABC_DJ has developed an algorithmic music recommender system specifically for use in audio branding scenarios. The ABC_DJ team first developed a validated list of marketing-relevant musical attributes comprising 36 terms (such as innovative, joyful, and authentic). Two international listening experiments followed, with a total of n=10,144 participants rating over 500 excerpts of music using these terms. The same pieces were then analysed on an acoustic and musical level using music information retrieval tools. Finally, machine learning was employed to map the correlations between acoustic features and ratings of semantic attributes. ABC_DJ furthermore addressed social variance in musical experience by incorporating values-based segmentation into the second listening experiment. The result is an algorithm capable of predicting perceived musical expression and liking based on the acoustic features of music pieces on the one hand and the characteristics and values of listeners on the other. In a commercial context, the resulting ABC_DJ system will allow brands to search digital music libraries in real-time, automatically updating their track pools at a pace commensurate with that of contemporary cultural production. The proposed presentation will demonstrate the ABC_DJ system. It will also introduce critical perspectives derived from qualitative interviews with n=23 expert users, who voiced attitudes ranging from scepticism in the aesthetic quality of algorithmically-curated playlists to fear that systems like ABC_DJ might devalue the labour of human music curators.

M3 - Conference contribution

BT - Algorithmic Music: Value, Creativity and Artificial Intelligence A One-day Symposium

ER -