Towards Developing Non-Conventional Forensic Musicology: Part I

1. What Constitutes a Musical Work

Any analysis of music copyright must begin with the definition of a musical work. The 1957 Copyright Act defines a musical work as ‘a work consisting of music’.[1] This definition may include notations of musical works, and excludes words and actions ‘intended to be [.…] performed’ with music. Notably, it does not define ‘music’ per se. However, since a musical work must ‘consist of music’, it is necessary to attempt to define music to understand and apply the definition in practical contexts. This exercise presents several challenges.

What comprises music is largely determined by the time^[2] and the culture within which it exists.^[3] Music resists definition; it would be unproductive, perhaps even impossible, in light of the total absence of scholarly consensus on the matter,[4] to provide a comprehensive definition of what ‘music’ is. Nonetheless, the law demands that the term be understood.

Music is frequently described as organised sound. In utilising this definition, we may limit our understanding of ‘organised sound’ to sound arranged in time according to relationships of pitch and rhythm (a ‘tonality’ requirement), in a manner that is perceived as meaningful or structured (an ‘aesthetic’ requirement).[5] This understanding, while intuitive, is imperfect.

The immediate problem with this definition is that there are types of organised sound that are not music, such as speech. Further, to exemplify the challenged posed by this definition: John Cage’s 4’33”, in which the performer produces no intentional sounds whatsoever has been presented as a musical work, and taken seriously as such by musicologists,[6] although it is not easy to convincingly argue that it could be considered to be a protectible musical work from the point of view of copyright law. If silence or ambient ‘noise’ can be music, then ‘organised sound’ probably does not tell us the full story.

A more accurate definition, perhaps, is that music is sound fashioned in time.[7] Accordingly, a musical work is a sound structure indicated (though it need not be definitively determined) by its composer. What makes up this sound structure has been extensively debated. The generally accepted constituent elements of a musical work in the Western world are pitch, rhythm, timbre, tone colour, dynamics and tempo which provide texture to music and structure it through melody and harmony.[8]

A musical work may not always be made up of all of these elements (Indian classical music, for example, generally excludes the aspect of harmony); any one element or a combination of them may also suffice to create as a musical work if it is, in some sense ‘sound fashioned in time’.

In assessing what makes up a musical work, certain elements can be excluded from our discussion. Philosophers of music have tended to confined themselves to analyses of ‘music alone’[9] which excludes non-musical aspects such as lyrics. This separation serves three functions: first, pure music presents its own set of unique philosophical problems, such as whether a piece of music alone (independent of its lyrical content) can have any emotionally expressive value at all; second, solutions to these problems are more easily evaluated by assessing music alone, and third, music’s distinct qualities affect our evaluations of composite works (like songs with lyrics).[10]

This functional separation of music and accompanying elements is also useful legally: it is simply easier to assess a musical work independently. Under copyright law, any semantic content associated with music is the concern of literary copyright, and attracts protections comparable to those relating to musical works but which function separately under Section 14(a) of the 1957 Copyright Act. Copyright in music is concerned with sound structure alone, which perhaps explains why the definition of a ‘musical work’ in Section 2(p) of the 1957 Copyright Act excludes ‘words [….] intended to be sung, spoken or performed with the music’. This categorisation is, however, tested by forms of vocal music that do not have a lyrical component such as melodic chants that exist in both the Western and Indian traditions where the voice functions in the same way an instrument would.[11] Such music would still fall within the statutory definition of musical works, which merely excludes ‘words’ from its scope rather than excluding all vocal music.[12]

All musical works are not immediately copyrightable. Where a musical work exists, it must be the expression of an idea, and it must be original in order to attract protection. Further, although a musical work need not be fixed with reference to its statutory definition, fixation is generally necessary for evidentiary purposes: even if an unfixed musical work is copyrightable, it would be impossible to prove the existence of such copyright (or indeed of the work itself!) without any record of the work. Each of these requirements are addressed subsequently.

2. The Fixation of Musical Works

For a musical work to exist as a fixed object, it must either be scored or recorded. By interpreting the statutory understandings of musical work and performances, it may be deduced that the current position under the 1957 Copyright Act (following the 1994 Amendment to the Act) is that the score, the performance, and the recording of the performance can exist as copyrightable works independent of each other. In particular, the score and the performance can exist as independent music works: neither requires the other to qualify as a musical work; a ‘performance’[13] could simultaneously be a ‘musical work’ under current law.

The initial definition of a ‘musical work’ in Section 2(p) of the statute carried a notation requirement prior to being amended in 1994;[14] a musical work had to be expressed in graphical form to attract protection.[15] The fixation requirement was essentially synonymous with a notation requirement. A performance alone did not fall within the statute’s definition of a ‘musical work’ if it was not graphically notated. This requirement was not especially problematic for works within the Western musical tradition, which uses a well-defined and universally-accepted notation system – Western music is capable (at the very least) of being graphically notated. The emphasis on the score as musical work confined this understanding to music that could essentially be expressed through Western musical notation. For Indian classical music, however, the notation requirement raised concerns, which led to the amendment of the definition.

Following the coming into force of the 1994 Amendment Act, notation is simply one possible fixed expression of a work rather than a precondition for it. The current definition is more aligned to a musician’s understanding of what a musical work actually consists of, particularly in the sphere of Indian music.

A musician finds it easy to understand the idea of a performance existing without notation; it is more difficult to conceive of a score alone being a work of music. More than any other copyrightable work, music is realised through performance. In utilising an understanding of music as sound fashioned in time, it is essential to acknowledge that music is inseparable from its temporal structure. Aesthetically, music derives its value from being heard and felt in time;[16] performance is the mechanism from which music derives its aesthetic value. As a result, a musician might find it difficult to conceive of a definition of music divorced from the aspect of performance.

Despite this, as indicated earlier, the law does not hesitate to separate the two. The legal definition of a musical work does not demand performance; a score is generally considered to be a musical work in and of itself. An ontological justification exists for this separation: a work is treated as a repeatable musical entity and the performance as one concrete realisation of it.[17] In that view, the work is not identical to any single performance, and performances are the events through which the work is heard and experienced.[18]

The distinction between the composer and the performer also merits consideration. Copyright law distinguishes performance from composition. The two are protected separately, the composer/composition through author’s rights, and the performer/performance via performers’ rights,[19] although, as previously discussed, it is possible for music to simultaneously be ‘performance’ and ‘musical work’ under the 1957 Copyright Act. While Western classical music tends to often be performed from a pre-composed score, genres like jazz are premised on improvisation by a performer. The distinction between performance and composition is not always clear-cut in practice, and may be better understood through two illustrative scenarios. 

In the first, a composer creates a work and a classically-trained performer subsequently interprets it during performance. The act of composition occurs independently of performance and interpretation although ‘interpretation’ cannot be separated from ‘performance’.[20] Therefore, the composer determines the ‘work-constitutive’ features of the sound structure which could be considered to be the essential features of the structure that give it its character as an original piece of music.[21] Within the latitude permitted by these work-constitutive features, the performer is free to interpret the piece by adding or integrating ‘interpretative’ details without disrupting the structure determined by the composer. Here, the lines between composer and performer are clear. In another sphere, this could be compared to an interior designer decorating a house without changing its structure, as distinguished from an architect renovating the structure.

In the second scenario, a jazz musician improvises to the extent of performing an essentially unique composition – the acts of composition and performance occur simultaneously, and cannot be meaningfully separated.[22] Even though such improvisation may be rooted in an existing work, there is so much deviation from the ‘work-constitutive features’ of that work that a new work emerges, in essence. In such a case, the improvisation itself, ought to constitute an original ‘musical work’, and be treated as a composition, not strictly a performance. The performer simultaneously occupies the role of a composer. In this second scenario, the performer essentially performs the role of an architect renovating an existing structure: the original structure remains but the final work could vary from it to a significant degree.

The situations described here do not, in themselves, create ambiguity. However, the performer-composer dichotomy is not always reducible to neatly distinct categories. It is useful, then, to provide a mechanism to distinguish the two scenarios in musical and legal terms which do not rely on analogy: to be considered a performance per se, firstly, the performance of a work (as opposed to an improvisation of it) must substantively align with the work in question; secondly, performers must intend to follow the instructions laid down by the composer; and thirdly, there must exist a causal chain between the work’s creation and the performance.[23] If these requirements are not met, a work ought to be eligible for treatment as a distinct work, that is to say, as an adaptation. 

3. The Expression of Musical Ideas

Copyright law also separates ideas, which are those generic concepts or formulations of thought that may be expressed in a number of ways, and expressions, which are the specific ways in which those ideas are expressed. The law grants protection only to the latter.[24] In assessing musical works, the ‘idea-expression dichotomy’ becomes important because most, if not all, music is built on a finite, and largely standard set of ‘building blocks’ that tend to remain constant within the musical traditions to which they belong.[25] These blocks could be considered to be the ideas that underlie musical works, and can be expressed in any number of ways. They may also be considered immutable fundamental units of music under the doctrines of scènes à faire (which deals with ‘scenes that must be done’) or merger (which recognises that ideas and expressions are sometimes so closely intertwined that they are inseparable). In all three contexts, they resist copyright.

Ideas, in this context, can also refer to either the standard themes and conventions that are ‘commonplace’ in the artform, or to the ‘profound or original ideas’ which may have initially been an original creation, but have been so widely adopted that they have become commonplace.

For example, the 12-bar blues progression (one of the most popular chord progressions in pop music; based on fixed sequences of the I, IV, and V chords of a key) might have constituted an original expression when it was first invented, but has been used so frequently in blues and rock music that it has ceased to be considered original. [26] Further, a musical structure or style, such as the four-movement sonata form developed during the Classical period is not copyrightable per se. As an illustration, if Beethoven’s 32 piano sonatas were composed today, they would almost certainly be protected as individual works. However, the fixed structure that all sonatas use cannot warrant protection without an expression.

Musical building blocks may include common melodic motifs, simple rhythms, or harmonic devices. Copyright theory, reflected in statute, disqualifies many of these commonplace conventions and compositional techniques from the scope of copyright protection. They tend to only qualify as protected musical works where they are formulated as passages that reflect a degree of originality. These aspects deserve independent consideration; they tend to make up the bulk of copyright litigation, and can be useful in determining what elements are not protectible, as courts may filter out unprotectible elements while evaluating originality.^[27] 

All Western music tends to be built on the same twelve notes, colloquially speaking. Each of these notes forms the basis of one ‘key’. The key of a piece of music is its basic musical point of reference and determines which of these twelve notes ‘make sense’ in the context of the music, and which notes don’t. Put simply, the key determines the set of notes (and chords) which sound ‘at home’ in the context of specific pieces of music. Usually, eight notes out of the twelve available notes will fall within a particular key; these notes can be used to construct melodies, harmonies and chord progressions.

Melody is a tonally and rhythmically organised series of sounds. In general, melody is the most distinct element of a musical work; it is made distinctive by variations in both pitch and rhythm.[28] While original melodic expression is certainly protectible, melodies are built from a limited musical vocabulary. Short melodic phrases, recurring note patterns (ostinato), and common scalar movements frequently recur across works. In Gray v. Hudson,[29] the United States’ Ninth Circuit Court held that repeating melodic motifs (ostinato) are commonplace musical elements, and the use of these motifs do not amount to infringement. The Delhi High Court quoted this decision in arriving at a similar conclusion in A.R. Rahman v. Ustad Faiyaz Wasifuddin Dagar.[30] The copyright protection afforded to such melodic devices is ‘thin’, and claims of substantial similarity on this basis will usually fail.

In the Western tradition, melodies are often supported by harmony, to which chords and intervals are integral. A chord is a combination of three or more notes played simultaneously with distinctive harmonic character. In every key, different chords have certain ‘functions’ within the key, and are used to create tension and resolution.[31] This effect (of tension and resolution) is achieved through chord progressions, which are sequences of chords played in a specific order, usually in a particular ‘harmonic rhythm’. The basic chord progressions are the tonic-dominant-tonic (I-V-I), tonic-subdominant-tonic (I-iv-I) and tonic-subdominant-dominant-tonic (I-IV-V-I). (Chord progressions are generally notated using Roman numerals. The tonic (I) refers to the first note in a scale, the dominant (V) is the fifth, the subdominant (IV) is, as the numeral suggests, the fourth and, finally, the submediant (vi) is the sixth.)

Some combinations and sequences of chords are more effective at achieving the tension-resolution effect than others. As a result, it is perhaps natural that, over time, certain chord progressions have emerged as ‘standard’. In pop music, certain formulae sell better than others[32] – as a result, almost all pop music is based on a handful of common chord progressions, which sound familiar to listeners, and which are generally evocative of certain emotions. A notable example of this is the use of the I-V-vi-IV chord sequence, which appears almost three times more often than any other chord progression in the Billboard Top 100.[33]

Chord progressions are usually self-contained, designed to repeat indefinitely, and can be initiated on any chord in the sequence using the same harmonic logic. The result is that a progression such as I-V-vi-IV is, in practice, interchangeable with IV-I-V-vi or vi-IV-I-V and identical in harmonic substance. A piece of music with this shared syntax does not necessarily cease to be original for having utilised it[34] but, equally, it cannot claim originality on the basis of the chord progression alone. As such, common chord progressions[35] do not appear to be protectible due to their standardisation nor do the ‘harmonic rhythms’ to which they belong seem to automatically cross the threshold for copyrightability. 

Harmony can also be achieved merely through harmonic intervals. Intervals refer to the inclusive distance between two notes or tones; if the notes are played simultaneously, the interval is a harmonic interval (in contrast to a melodic interval where notes are played successively).

Harmony is, simply put, the sound of two or more notes heard simultaneously. Different combinations of notes including harmonic intervals and chord progressions produce different ‘sounds’. These sounds may be consonant or dissonant. Certain (consonant) harmonies are more pleasing to the ear than others,[36] based both on the nature of the interval, as well as their context within a musical work. Harmonies serve a functional purpose in musical works. They ground a melody in a particular key, provide the bases on which chord progressions are built, and allow listeners to distinguish the character of the music, in particular, the melody. 

Harmony, arguably, cannot be copyrighted for three reasons. Firstly, a harmony may not have any musical value independent of the melody it exists as an accompaniment to. Without a melody, there is no real question of the existence of a harmony. Secondly, because musical intervals are standard in Western music, all harmonies are, in a sense, standard. Finally, harmonies have existed in Western music for centuries, and it is often near impossible to prove that a harmony is not already in the public domain.[37] As the ‘building blocks’ that underlie all musical works, common melodic motifs, chord progressions, and harmonies are likely to be considered unprotectible elements under copyright law.

4. Originality and Authorship in the Face of New Technologies

Copyright can only subsist in a musical work if it is original.[38] Until very recently, our understanding of originality has been intrinsically tied to human creativity – the present copyright law regime is broadly designed to protect, incentivise and reward this creativity. However, the widespread integration of generative artificial intelligence (‘AI’) across nearly every stage of the creative process^[39] is already raising questions about whether (and if ‘yes’, how) the law is equipped to assess originality in a world where machines are capable of performing tasks that formerly required human or human-like intelligence or creativity. AI has also begun to pose a distinct challenge to copyright law’s conception of authorship: where does it vest when a musical work is brought into being using AI?

An approach that extends copyright protection to AI-generated works would, in some sense, favour ‘the proliferation of creative works in the hands of consumers’[40] regardless of whether their origin was human or artificial, potentially facilitating greater accessibility of creative works. The Berne Convention does not provide much guidance on the matter, and different copyright regimes have interpreted the originality requirement in the face of generative AI technologies differently: the United States Supreme Court[41] and the Court of Justice of the European Union[42] have both supported the position that originality requires some level of human creativity. Chinese courts, on the other hand, have granted protection to AI-generated works, if other requirements are met.[43]

The Department-Related Parliamentary Standing Committee on Commerce, in its Review of the Intellectual Property Rights Regime in India recommended incorporating AI-related developments within the ambit of the Copyright Act.^[44] In 2020, Indian Copyright Office granted registration to RAGHAV, an (artistic) work that was jointly authored by an AI model and its programmer;[45] the registration of the work was later withdrawn. In Canada, the same work was permitted to be registered, while the US Copyright Office rejected the application, citing a lack of human authorship. 

While the 1957 Copyright Act (unlike some other copyright regimes) does not explicitly require ‘human’ authorship,[46] necessarily, it does require originality of musical works, and demand that sound recordings be non-infringing to qualify for copyright protection. At present, the statute lacks a clear mechanism for attributing authorship where the specifically musical choices were made by a system and not by any identifiable person. 

The idea of a derivative work lying somewhere along a ‘spectrum’ of originality (discussed in detail the following section) may prove useful as an illustration. At the mechanical end of this spectrum, a human performs a musical function with no creative input (such as notating a work using MIDI software). At the opposite end, generative AI may ‘create’ musical works involving harmonisation, arrangement, orchestration, etc., with minimal human input. At both ends of this spectrum, the problem is the same: our idea of ‘originality’ is challenged. In the first situation (the ‘mechanical end’), the ‘modicum of creativity’ test may provide some guidance: where some amount of skill or judgement is exercised by the author, a work is eligible for protection; in its absence, no protection is warranted.

In the second situation, however, the law is not currently equipped to assess originality. At present, generative AI’s output, no matter how technically impressive, is only capable of producing statistically coherent outputs based on training data. There is some distinction between music generation, which involves the production of an ‘instance’ of music within a given style or framework, and the creative exercise of making music, which requires engagement with concepts that often lie beyond the notes and sounds.[47] Applying the same ‘modicum of creativity’ standard, it is difficult to say whether a work entirely created by generative AI (even if broadly conceptualised by a human) could be considered truly “original”.

Any assessment of AI-generated work is complicated by the involvement of a human prompter. When a work is eligible for protection, the author is usually the first owner of the copyright therein;^[48] they may subsequently assign that copyright to a third person.^[49] Section 2(d)(vi) of the 1957 Copyright Act vests authorship of computer-generated works in ‘the person who causes the work to be created’.[50] This provision was drafted in 1994 when computers were understood as tools directed by human operators, and implies a causal link between the author’s input and the creation of the work. In ascertaining the existence of this causal link, a degree of specificity must be established between the output and the instructions provided by the human prompter. This means that for an AI-generated work’s authorship to vest with the human who provided the prompt, the output must reflect the contents of the prompt with specificity, not merely in a general sense.[51] Specificity implies originality (and is more useful in determining infringement), while generality suggests a lack thereof. The present provision serves its purpose where AI is used as an ‘assistive tool’, that is, as a compositional aid alone as specific intent can still be attributed to the author. 

When AI is used as a generative tool (as opposed to an assistive tool), authorship is harder to ascribe. Technically, a prompt by a human is sufficient to cause a work to be created. Current music-generation systems respond to broad-level prompts (on musical attributes such as genre, style, instrumentation, key, duration); it is nearly impossible to ascertain the degree of causation between the composer’s instructions and the final (AI-generated) composition. AI also suffers from a ‘black box problem’, which makes it near impossible for humans to ascertain how deep learning systems actually arrive at their outputs.[52] 

As a result, it is nearly impossible to establish the existence of a causal link between authorial intent and the actual creation of the work. A generic prompt (such as “a sad love song” or “an upbeat rock song”), which merely provides a list of general musical attributes that the final work must possess, may produce any number of statistically coherent results, based on how AI interprets the prompt. These prompts are more in the nature of ideas, not concrete expressions, and would likely not merit protection on the basis of the idea-expression dichotomy. The expression, if any, can only be attributed to the AI tool, not to the human who provided the idea. As discussed, AI tools themselves are not strictly capable of originality (within our current conception of the word). Due to this, a work originating from a non-specific broad-level prompt is unlikely to be considered protectible. 

The considerations listed above are part of a broader discussion on whether copyright law’s current conceptions of authorship are sufficient to encompass what creativity will look like in the future. Authorship, despite its centrality in copyright law, is a ‘shifting concept that is sensitive to context.’.[53] Until the 20^th century, writing music (in the Western classical tradition) was a largely solitary activity; the author was largely synonymous with the composer of a work. This idea translated into statute.^[54] The developments of the 20^th century made the process of creating music far more collaborative, and it became commonplace for multiple people to be involved in the creation of a work. In many ways, these developments also democratised the process of creating music, separating the requirement for formal musical training from the actual creation of a musical work. High-level composition no longer required the ability to read, write, or play music.^[55] Our classical conceptions of authorship were nevertheless able to accommodate this shift. Despite this, they remained tied to an understanding of human creativity (as a product of the intellect of the creator) that was developed in the 16^th century, and has changed very little since.[56]

21^st-century technologies are likely to alter the relationship between the composer and their work so significantly that our ideas of authorship (and creativity) may need to expand if we are to adequately protect musical works in the future.

In the meantime, a specificity threshold is perhaps worth incorporating in the present legislation. A prompt which ‘causes’ a work to be created must be specific enough to establish authorial intent to create that work; it cannot be generic to the extent that the same provided parameters could give rise to a substantially different work. It may become necessary for the law to assess the process behind the creation of the work in order to determine whether authorship can be established. 

5. Adaptations and Copyrightability

Both original and derivative works are protectible under the Indian copyright statute provided they do not infringe on an existing copyrighted work.[57] The Act contains specific provisions which relate to one form of derivative work: adaptations. Sound recordings, which are also sometimes derivative musical works, are protected separately under Section 13 of the 1957 Copyright Act.[58]

Adaptations are generally understood as works which alter an original work in some way while retaining the core distinguishing qualities of the original work. Section 2(a)(iv) of the 1957 Copyright Act defines ‘adaptation’ in relation to a musical work to include ‘any arrangement or transcription of the work’.[59] In addition, any use of a work involving its rearrangement or alteration of a work is also considered an adaptation under Section 2(a)(v).[60]