Since the invention of the phonograph in the late 19th century, music has gradually evolved from an art form and folk culture into a domain of popular culture, eventually forming today’s music industry, which has generated remarkable economic value. From a legal perspective, because music creation exhibits the characteristics of human intellectual creation, it is classified as cultural intellectual property and is primarily protected by copyright law, enjoying various copyrights guaranteed by copyright law.
The music industry’s protection under copyright law is primarily manifested in three types of works: musical works, sound recordings, and performances. According to Taiwan’s Copyright Act and related regulations, musical works include sheet music, lyrics, and other musical compositions, with primary products being lyrics and songs. Sound recordings include any works featuring a series of sounds expressed through mechanical or electronic devices that can be fixed on any medium—ranging from vinyl records, cassettes, and optical discs to digital audio files in the internet age. Regarding performances, the Copyright Act explicitly stipulates that performers’ performances of existing works or folk creations should be protected as independent works.
Over more than a century, the music industry, on the basis of the protection provided by the above copyright system, has largely managed through successive technological innovations and challenges to achieve successful transformations. Particularly, the wave of piracy brought by internet proliferation, though completely reshaping the industry’s ecosystem, did not fundamentally shake the core products of the music industry and their value—musical works, sound recordings, and performances.
AI is Bringing Major Transformation
However, since ChatGPT’s emergence and entry into public life, AI’s impact on various industries has surfaced. Generative AI possesses distinctive characteristics of “self-generation” and “creativity” different from other digital tools, capable of independently conducting creative activities in autonomous states and producing various works. This leaves creative industry professionals unable to resist asking themselves: “Will my job be replaced?”
Beyond ChatGPT’s impact on text and knowledge industry workers and Midjourney’s impact on visual artists, the music industry similarly has extensive relevant discussions. So from a copyright perspective, what impact will AI have on the existing copyright protection system for music creation, and how might it affect the music industry?
The Birth of a Song
Let us first understand how a song is created. According to the current production process common in the popular music industry, a song typically first undergoes lyric and melody composition, then arrangement, or alternatively, background music is created first, followed by lyrics and melody. Subsequently, after creating a demo, recording sessions are held to record all or part of vocals and instruments as needed. Finally, the song is completed through mixing and mastering steps.
The roles involved in these processes can be roughly divided into:
- Music Creators: Individuals providing melody, lyrics, harmony, rhythm, and other musical compositional elements, such as lyricists, composers, arrangers, and beatmakers.
- Music Producers: Individuals making actual contributions to sound elements and sound effect presentation in audio-formatted music, such as producers, recording engineers, mixing engineers, and mastering engineers.
- Music Performers: Individuals expressing music through different interpretative methods, techniques, and forms of expression, including singers and musicians.
Different roles make different types of contributions to a song, and these contributions are all “intangible assets created by human intellect in the artistic realm.” Therefore, they also legally enjoy different types of copyrights—musical works, sound recordings, and performances.
Legal Requirements and Rights Content of Copyright
Additionally, for music creation results to receive copyright protection, there is an important prerequisite: confirming whether the object to be protected constitutes an “work” as defined by the Copyright Act. The recognition standards are four:
- Must be the result of human intellectual activity;
- Must be externalized through “expression”;
- Must be independently created and possess creativity;
- Must belong to the realm of literature, science, art, or other academic fields.
Regardless of which type of music creation result, it must first be correlated to the above criteria to become a “work” protected by copyright law. After confirming it is a “work,” output results in different forms of expression are classified legally into different types of works:
- Musical compositional elements constitute “musical works.” For example, melodies (composition), lyrics (words), and sheet music.
- Audible sound recordings constitute “sound recordings.” For example, analog-signal-recorded audio such as records and cassettes, or digital-signal-recorded audio such as digital files.
- Musical expressions constitute “performances.” For example, live band performances or live singing by singers.
After these types of works are created, to whom does their copyright belong? According to copyright law provisions, in general circumstances without other legal provisions or contractual agreements, “music creators” enjoy rights to “musical works,” “music producers” enjoy rights to “sound recordings,” and “music performers” enjoy rights to “performances.” The content of these rights includes the vast majority of uses that can promote the dissemination, circulation, transactions, earning, and realization of works. The Copyright Act collectively refers to these as “copyright economic rights.” These rights include controlling others’ use, distribution, or modification of works and often represent the basis for revenue distribution.
AI’s Impact on Popular Music Creation and Industry Professionals
Since copyright economic rights affect revenue distribution in the music industry, how might AI’s involvement potentially impact industry professionals?
Currently, AI applications related to music already include the following tools or forms:
- Using AI to generate lyrics, melodies, beats, and other musical compositional elements.
- Using AI voice synthesis to replace actual recording results or create sound effects different from recordings.
- Using AI mixing and mastering tools.
- AI voice synthesis or imitation of specific singers.
Some of these may be tools to assist creation, but others directly replace elements of human creation, potentially causing creation results to lack the requirement of “being the result of human intellectual activity,” making them unable to constitute “works” and thus not protected by copyright law. Music industry professionals find it increasingly difficult to maintain the status of rights holders and enjoy revenue from such results.
Music’s Core Rights Facing Impact
For example, in using AI for mastering (final audio processing), mastering engineers currently still invest technology and experience in the work, and the related mastering tools are controlled and operated by producers and engineers. Therefore, such work still meets the requirements of human creation and can in principle constitute protected works enjoying copyright protection. However, if completely using fully automated mastering tools, whether such mastering work results can enjoy copyright is questionable.
For another example, using AI music creation tools where users only provide themes and keywords, with AI directly generating lyrics, melodies, beats, and other elements, if humans do not participate in the creation process of these musical compositional elements, such AI works very likely would be determined to lack sufficient human intellectual input and thus do not constitute legally protected works.
Recently, we can also observe the emergence of numerous AI voice models. AI voice models, by allowing the model to analyze and learn human voice audio data, establish independent singing capabilities. Whether inputting existing songs to AI voice models for the model to sing constitutes “performance” as termed by copyright law is questionable. Furthermore, if the AI model is trained based on a specific singer, can that specific singer claim any rights?
Music Industry Professionals Facing Impact
From the three examples above, we can discover that in today’s age of AI, the core music rights traditionally constituted by musical works, sound recordings, and performances are facing significant impact. As music workers increasingly become accustomed to using AI tools, while possibly greatly improving work efficiency, speed, and output, they may simultaneously weaken their legal status as industry professionals.
When music industry professionals work in traditional ways, their output constitutes musical works, sound recordings, and performances. They enjoy the economic value brought by their work as creators and copyright economic rights holders, benefiting from the rights system protection constructed by copyright law. If music industry professionals become operators of AI tools, they will inevitably lose the above legal status, devolving into assembly line workers in music production, becoming knowledge sector blue-collar workers exchanging labor for wages.
Additionally, even if some professionals adhere to traditional creation methods, allowing their works to still enjoy copyright protection, due to the prevalence of AI tools, the threshold for producing music becomes lower than before. This means commercial copyright music may be obtainable at even cheaper prices. Considering costs, music users will inevitably abandon expensive copyright music (simultaneously avoiding potential copyright disputes) and instead widely adopt AI-created works without copyright protection. Consequently, the commercial music market may be occupied by AI-created music, crowding out music creators’ income and making the survival space for copyrighted music even narrower.
Conclusion
AI has become a prominent field of the 2020s in the 21st century. Whether AI will ultimately bring humanity a science-fiction-style apocalypse or a utopian beautiful new world remains perhaps too early to say. However, AI has indeed begun transforming human learning, work, and life in many respects, bringing innovation, opportunities, and beautiful expectations, while also bringing imaginings of challenges, risks, and fears.
In the future world with AI, music, the music industry, and its professionals, like other fields, will face both opportunities and challenges. But a more severe scenario is that the legal protective umbrella on which the music industry has relied for over a century—copyright protection—will itself be shaken. What this article discusses is merely the first few measures in the grand prelude to this enormous transformation. The main melody is just beginning. How industry and law will respond to the future, we hope that “human intelligence” will ultimately surpass “artificial intelligence” by a small margin.
