Listen Local Collaboration Guide
2022-07-15
Short Introduction
Listen Local is an open collaboration for artists, fans, managers, developers to provide alternatives to connect local audiences with local artists through locally relevant music recommendations and findings. As a project of the Digital Music Observatory it aims to make big data work for small labels and self-released artists, and to make algorithms work for and not against them.
Listen Local is system of trustworthy AI applications developed and tested in a transparent community, using open source algorithms, open data, and open source applications. The “trustworthy AI” concept is the European Union’s regulatory concept to make sure that that AI algorithms work in an ethical and transparent way. In the context of music, a trustworthy algorithm helps womxn, or at least is not biased against them. A trustworthy algorithm does not contribute to the colonization of music ecosystem of Vilnius or Gent with Northern and Latin American music, and makes sure that local users, particularly B2B users like shops and restaurants leave some visibility and royalty with the local creators and labels.
Figure 0.1: The planned service flow of the collaboration.
Listen Local is developed by a community under managerial support of Reprex, a Dutch start-up that is part of the EU AI Alliance and Dutch AI Coalition. The aim of these initiatives is to make big data and AI work for all, not only the big corporations and powerful institutions. It relies on a broader open collaboration among universities, SMEs and CSOs, the Digital Music Observatory, that aims to help data collection and improve the reliability of the data. Read more about our ► Digital Music Observatory.
This Collaboration Manual is created for the existing and future contributors of this project. You can download the latest version of the manual from listen-local-collaboration.dataobservatory.eu. The last authoritative copy of this manual (Version 2022-07-15) is doi.org/10.5281/zenodo.6617137.
Ethical, trustworthy applications
Most of our application ideas use some form of automation or artificial intelligence. We create smart application (see ideas below). To make these applications, Listen Local’s app must use machine learning from a reliable information. YouTube or Spotify are among the most developed users of machine learning and its heavy-weight new version, deep learning. We will never compete with them in algorithm design and engineering. We offer an alternative where a community can create more value than even more complex AI systems, big data, and billions of iterations. We make sure that our algorithms—and their algorithms—are learning from clean data.
Any AI system is as good as the data that goes into it. Big data works against womxn, against small countries, against small and independent labels because the proprietary commercial AI learns from data that was created by patriarchal, institutionally racist datasets created by commercial interests. Listen Local works with ethical, transparent, clean data that is free of biases based on age, body size, visible or invisible disability, ethnicity, sex characteristics, gender identity and expression, level of experience, education, socio-economic status, nationality, personal appearance, race, caste, color, religion, or sexual identity and orientation1.
The Listen Local Slovakia project created a feasibility study with rich demonstration application to prove our concept with the support of the Slovak Arts Council. The next step, Listen Local Lithuania aims at creating the first, replicable, community-based, affordable trustworthy applications and services.
Read more about our ► goals, project ► history, various ► application ideas. most of this is very sketchy.
Trustworthy Information
All good music should find its audience. Local ecosystems, such as shops, events must find ways to play local music, introduce visitors to local artists, and make sure that some of the royalty payments (for example, paid by restaurants or bars) stays in the local music ecosystem. When somebody is shopping in Milan for fashion, in the shops some music from Milan and its regions should be played.
Listen Local facilitates the best human-human and human-computer interactions, and makes sure that machine learning algorithms do not learn from bad information, leading to bad recommendations, and unsuccessful artist placements and low valuation and appreciation for their music.
The Listen Local Databases are location-specific and subjective databases about artists and their works. Listen Local Slovakia contains artists and music references that are relevant in the context of Slovakia: the music is made in Slovakia, or the artists or music relates to the Slovak language and culture. Listen Local Gent is music relevant in the city of Gent.
6 The Write-In Databases contain only publicly and legally available information collected and organized by the Listen Local Collaboration with the oversight of local music curators. The write-in databases contains information that our automated systems with the oversight of our curators write into them. For example, the Slovak Comprehensive Music Database contains music that our Slovak music curators subjectively think to be Slovak music in one way or another. Our curators use a sound and consistent set of criteria to write in music and artists into our databases, but they maintain their subjective and professional judgement.
7 The Opt-In Databases contain information that creator of the music provided about herself. The opt-in databases consist of two datasets: a closed and GDPR protected personal identification dataset and a public dataset that contains information released with the consent of the creator. Any artists can opt into our databases pending curatorial approval. (An artist can opt-in into the Slovak Comprehensive Database if he or she feels Slovak, pending the judgement of the curator that the inclusion of the artists will be relevant for music audiences and users looking for Slovak music.) Any artist can opt-out from the personal dataset without curatorial approval.
7.1 Any artist can opt-out from the personal dataset without curatorial approval. Special rules apply on deceased artists—partly, because GDPR no longer applies, partly, because identity will be relevant for new generations. For example, the classical work of Franz Liszt lived and worked in the Austro-Hungarian Empire, and his work may be credibly claimed by the audiences and researchers of Slovak (he worked much in the current Slovak capital of Bratislava), Hungarian (his place of birth) or the music heritage of the city of Weimar in Germany. The inclusion of his work Weimar, Hungary (very likely) or Japan (unlikely) specific datasets requires the judgement of our curators.
The principles of the Listen Local system were outlined in the Feasibility Study On Promoting Slovak Music In Slovakia & Abroad. This study was supported by the Slovak Arts Council and the state51 music group, and it is avaialble in both English and Slovak languages2.
Participate
Listen Local is a highly automated system that uses open-source algorithms for automation and open, trustworthy AI. We deploy these technologies to help humans to work with music, enjoy music and participate in music. Keeping in mind a basic principle of ethical, trustworthy AI, our system is always maintaining human agency and oversight.
- 2 Curators are the most important collaborators to make Listen Local human-centered and ethical. They manage the write-in database and ensure that the automated systems collect and store correct information about artists and their music. To maintain ongoing data ownership of artists, they manage the opt-in and opt-out procedures for artists.
References
All our contributors must abide by the Contributor Covernant Code of Conduct which is maintained by the open source community that pledges to be fully inclusive for creators and developers.↩︎
Slovak version: (Antal 2020b), English version: (Antal 2020a).↩︎