Close

Artificial Intelligence, Hindu Art and Aesthetics


Abstract

Although aesthetics is studied from an audience perspective, it is evident that an aesthetic experience is evoked due to certain structural characteristics of the works of art. This involves careful and calculated craftsmanship of an artist or a creator based on certain domain-specific canons of art. In the realm of Indian literature, theater, architecture, sculpture, painting and music such canons have been rigorously theorized and practiced. This paper/presentation attempts to discuss the computational facets of these canons and show that such computational orientation of Indian arts provides strong philosophical foundations for the possibility of artificial creations as well as the legitimacy of aesthetic experience evoked. Based on the canons of music and aesthetics from Natyashastra and Sangeet Ratnakar, the author has developed a creative artificial intelligence system to generate a Raga Bandish that is showing promising results in this regard.

Introduction

In the West, the term aesthetics has genesis in Greek philosophy but the term has been widely used in recent time for last two centuries. Mostly this term refers to the person’s response to a beautiful object, whether natural or man-made. [Osborne Harold (Ed.), 1975, p.-12] and [Urmson, J. O., 1989, p.-2]. However, in India the term was not used in this restrictive sense. From very early theories of poetics in India it was always theorized that the phenomenon of aesthetic experience is a wholistic process. The theory of ‘Rasa’ discussed in Bharata’s Natyashastra always envisioned the process of aesthetic experience as a continuum involving, the creator, the work of art and the audience. In poetics and theories of language, the underlying principle of aesthetic experience is considered as the principle of ‘Pratibha’. Linguists and aestheticians such as Bhartrihari, Anandavardhana, or Abhinavagupta [Kuanpoonpol, Priyavat, 1993], give due importance to ‘Pratibha’ and its role in the process of aesthetic experience. Pratibha is responsible for creation of the work of art as well as reception and perception of the work of art. According to Rajasekhara, ‘Karayatri Pratibha’ is responsible for creation of a work of art and the ‘Bhavayatri Pratibha’ is responsible for its appreciation. [Rajasekhara, Kavyamimansa, Dalal, C. D. (Ed.) et al, 1934, Chapter IV, p.-15].

There exists a significant amount of literature on the appreciation process of aesthetic experiences, however, there is scanty literature about the factors that make a work of art capable of evoking an aesthetic experience. The creator brings in artistic elements based on certain definite principles that cause the evocation of an aesthetic experience. The common audience may not be aware about these elements and principles but the well-informed audience or the connoisseur is well-aware about these aspects. This paper attempts to discuss these creational aspects of the work of art that evoke the aesthetic experience. These involve careful and calculated craftsmanship of an artist or a creator based on domain-specific canons of art. In the realm of Indian literature, theater, architecture, sculpture, painting and music, such canons have been rigorously theorized and practiced. These theories have strong philosophical foundations. Vedic and Upanishadic frameworks have shaped up the artistic activities that lead to creation of aesthetic experiences.

Philosophical Foundations

असद्वाइदमग्रआसीत्।ततोवैसदजायत।तदात्मान्ँस्वयमकुरुत।तस्मात्तत्सुकृतमुच्यतइति।यद्वैतत्सुकृतम्।रसोवैसः।

[Jog, D, V. (Ed), 2018, pp. 221-222]

This may be translated as ‘Non-Existence’ or ‘That’ existed then. There emerged ‘Existence’. ‘That’ created itself, and therefore, it was ‘Sukruta’– the ‘Well-Made’. The ‘Sukruta’ is ‘Rasa’. These profound words from Taittiriya Upanishad provide philosophical context to all the later discussions on aesthetics in India. The term ‘Rasa’, which is understood as aesthetic emotion/experience in the domains of poetics, literature and theatre has a legacy in these Upanishadic shlokas. It says that there existed an ‘indescribable’ entity or an entity that is ‘beyond qualities and so cannot be described by words ‘(Gunaateeta Brahman). It created/transformed itself in the form that has qualities (SagunaBrahaman) and that is describable in words. Since the Brahman has created itself (Swayam krutah) the SagunaBrahaman is called ‘Sukruta’, – ‘The Good Deed’ or the ‘Holy Act’.

In the domain of artistic creation, this cosmic/metaphysical phenomenon is interpreted and metaphorically associated with the creative activities of an artist. Thus, it is assumed that an artistic ‘Creation’ is the ‘Creator’ itself. A creator transforms his or her ‘being’ into a ‘creation’ or a work of art. The world perceives it as the artist’s expression – Art as ‘Self-Expression’. In this sense human creation is analogous to cosmic creation and the phenomenon of cosmic creation may be correlated to human creation.

Art and Aesthetic Experience

Taking the analogy forward, various aspects of cosmic creation can be correlated to human creation. The above-mentioned Upanishadic shloka says that the cosmic creation is called ‘Sukruta’, and it is nothing but ‘Rasa’ itself. The ‘Rasa’ leads to ‘Ananda’, the ‘Cosmic Bliss’. Similarly, the human creation leads to aesthetic bliss or aesthetic experience. The above-mentioned Upanishadic thought is further elaborated in Vedanta philosophy. From the philosophical point of view of Vedanta, aesthetic activity and aesthetic experience is understood as described by a noted Indian philosopher Hiriyanna as follows. In Vedanta, Brahman– the Ultimate Truth – represents the ‘Inner Harmony’ of the Universe.  Realization of Brahman is the realization of Universal Harmony. A metaphysical goal of every artist is to attain this state which is direct and not mediated. As per Vedanta, the aim of art is to create an ideal work of art by inducing an attitude of complete detachment and try to be as close as possible to the state of Brahman. Therefore, the work of art should lead the artist as well as the audience to that restful bliss, realization of that harmony. Universal harmony is realized in one’s own experience and not merely intellectually apprehended (Hiriyanna 1954: 6–10).

Thus, the ultimate objective of artistic expression is to evoke a unique kind of aesthetic experience – take the audience into the realm of non-worldly, virtual experience. Artistic creation therefore is a complete world in itself. To achieve this, the artist constructs an imaginative world that is complete by itself. A painting, a sculpture, music performance or a theatrical performance is a complete world by itself. A performing artist creates an imaginative world using the artistic devices of language, poetic phrases, dialogues, rhyming words, beautiful hand gestures, and body postures. In painting, sculpture, and architecture, artists use the visual language of shapes, forms, colours, texture, positive and negative spaces to create reality that is complete in itself. This artistic world has its own logic, its own purpose and its own personality. The creators’ thoughts and actions are reflected and manifested in this reality. The creator has the total command over this world. Anandvardhana in ‘Dhwanyaloka’ while elaborating on one of his ‘karikas’ articulates this thought as follows.

अपारेकाव्यसंसारेकविरेक: प्रजापतिः।यथास्मैरोचते विश्वं तथा वै परिवर्तते ॥[Anandavardhan, p. 312].

“In the boundless world of poetry, poet is the Prajapati (Brahma – the ‘Creator’). He changes the universe as and when he thinks the best.”Thus, the creator is the master of his creations or his artistic world. Therefore, an artist’s ability to visualize such a reality and his/her skill to manifest it authentically plays a vital role in this process. The process has a structure, logic and language of its own. There exists a language of creative expression and it requires a grammar of aesthetics. This grammar is implicitly present and is followed by artists intuitively. Artists also have their own explicit grammar or canons for the language of artistic articulation. Traditional treatises on art such as Natyashastra or Shilpashastra give elaborate description of the respective domain-specific grammars of artistic manifestations and the rules, canons and artistic conventions.

Art as Computation

Art is an act of creating, expressing, and making. Art is a presentation and representation of ideas, thoughts, feelings, and experiences. The word for art in Indian tradition is कला. It is etymologically derived from the root ‘kal’ which means counting, calculating, or computation. It also means to do, to make, to accomplish, to observe, to perceive, to recognize, or to take notice. The word ‘कल्thus covers a large set of activities. Artistic activity involves creative perception and calculation for visualization and articulation respectively. This etymological meaning of the word ‘कल्i.e., to calculate or measure led to further exposition and development of quantitative standards for artistic practices in India.

In the field of visual and performing arts, the calculative and computational aspects of art have received an elaborate exposition. Indian arts, known for their spiritual and metaphysical content are very much quantitative in their physical manifestations. This is aimed at achieving the aforementioned ‘Universal Harmony’. It is fascinating to see that all the theoretical concepts of Indian arts have either quantitative or computational aspects. Many of us are familiar with the idea of Tala-Mana Pramana ratios used in Shilpashastra. Similarly, all the Indian art forms have their own domain-specific measurement systems and computational grammars. It is beyond the scope of this paper to give an elaborate exposition to all the concepts from all the domains from a computational point of view.

Artistic creations have profound computational foundations in Indian arts. The computational foundations enhance the quality of aesthetic experience. They also facilitate the process of attaining the ‘Universal Harmony’. There exists a correlation between artistic computation and the aesthetic experience. In this paper this is demonstrated by analyses of concepts from Indian music. It is fascinating to see that underneath the evocative ability of Indian music there are quantitative and computational principles or a ‘Computational Grammar of Music’ that make this aesthetic experience possible. It is interesting to see how these rules that are computational in nature play an active role and contribute in generation of music that leads to an aesthetic experience. Bharata’s Natyashastra, Narada’s NaradiyaShiksha and Sharangadeva’s Sangita Ratnakar are well-known foundational treatises on Indian music. The principles and rules of the grammar of Indian music are well-articulated in these treatises. The following discussion is based on these treatises. To understand these rules authentically, the author of this paper has designed and made a special musical instrument called ‘Bharat Veena’. All the concepts from these treatises can be tested and validated on this Veena. Bharat Veena is designed and developed based on the description from Bharata’s Natyashastra.

Bharat Veena

  • Demonstration of the ancient musical theoretical concepts such as Shruti Nidarshanam experiment from Bharata’s Natyashastra is possible on Bharat Veena. This experiment is also known as `Sarana-ChatushtayiPrayoga’. It can be performed on Bharat Veena.
  • Bharat Veena has dedicated strings for Shadja Grama, Madhyama Grama and Gandhara Grama
  • Gramantara experiment mentioned in Natyashastra also can be demonstrated on Bharat Veena.

In following paragraphs, the readers will understand that the rules of the grammar of Indian music have computational qualities and how their manifestation leads to generation of musical concepts. It also explains how musical compositions lead to aesthetic experience. An important aspect of this discussion is that there exists an inherent generative logic foundational to these rules. A unique characteristic of the generative grammar is that from one base concept of Shadjaswara, by applying a few basic rules, all the other musical concepts can be generated. This is precisely the reason it is termed as a ‘generative grammar’. While discussing the nature of ‘Brahman’ and the evolution of this world, Upanishads have mentioned similar thoughts. There exists an inherent generative logic in the process of how the singular ‘Brahman’ transformed itself into the plural world. Especially, the Taittiriya and Chhandogya Upanishads have very convincingly described this process. They say, initially, the ‘Brahman’ was one. It decided to become many.

सोऽकामयत बहु स्यां प्रजायेयेति|[Taittiriya Upanishad, Brahmavalli- Anuvak-6, p. 218],

That Brahman desired that it will transform into many.

तदैक्षत बहु स्यां प्रजायेयेति|[Chhandogya Upanishad, 6.2.3, p. 420],

That Brahman envisioned or envisaged that it will transform into many.

In the spirit of above statements from Upanishads it is shown, in the context of Indian music, that the single base Shadjaswara or the ‘Aadhar’ swara, generates rest of the other swaras of a Saptaka and then within a Saptaka various musically significant relationships among swaras are established. All this is achieved by following certain musical rules inherently present. Eventually, all these relationships contribute to the generation of music. From just one single Shadjaswara the whole world of music, with all its multiplicities can be created. This phenomenon is demonstrated here in a step-by-step manner in the following paragraphs.

It is shown that calculation and computation lead to generation of Swaras and Shrutis of Indian music from the base Shadjaswara which is also called Aadhar Swara. Consequently, computable relationships between and among swaras led to generation of Jatis, Moorchchhanas, Taanas and eventually – Ragas. This underlying computational generative process results into beautiful music that has aesthetic qualities.

This can be demonstrated on a specially designed Veena called the ‘Bharat Veena’ as follows.

Swara-Sthapana & Shruti Generation

The entire length of the string of a Bharat Veena produces the natural Sa (Shadja) swara as shown in the following figure.

As a next step, higher Shadjaswara is located exactly on the mid-point of the Veena. Following the same ratio all the higher Saptakas can be established as shown in the following figure.

व्यवहारे त्वसौ त्रेधा हृदि मन्द्रोऽभिधीयते| कण्ठे मध्यो मूर्ध्नि तारो द्विगुणश्चोत्तर:|[Sangita Ratnakar 1-3-7]

The next step is of establishing other swaras. Madhyamaswara is exactly on the mid-point between Shadja and the higher Shadja swaras. Since it is located on the midpoint, it is called `Madhyamaswara.  The Panchama swara is located on the point having the ratio 2/3 between Shadja and the higher Shadjaswaraas shown below.

In all the above figures it is shown that the entire length of a Veena represents a Shadjaswara. However, it is not always necessary. Any point on the string can be considered as a Shadjaswara or the Aadharswara and then by applying the rules described in Bharata’s Natyashastra it is possible to establish a Saptaka and then twenty-two shrutis can be established on the Bharata Veena. Bharata’s system was different from contemporary system. It was established on equal temperament shruti paradigm. Bharata Veena is a ‘Sama-ShrutiVeena. Therefore, the value of a ‘Pramana Shruti’ is 1.032 ratio which is validated by the ‘Shruti-Nidarshanam’ experiment mentioned in the Natyashastra. The Shadja-Madhyama Bhava and the Shadja-Panchama Bhava are also based on this Pramana Shruti ratio. Bharata’s Saptaka and twenty-two shrutis are also based on this ratio. In the following paragraphs, the method for establishing a Bharata’s Saptaka is discussed and then shruti generation method is also discussed later on.

Following paragraphs explain the whole process in a step-by-step manner.

Establishing Saptakaon Bharata Veena

First establish a Vadi swara or the Aadharswara (Bharata calls the Aadharswara as Vadi because it is useful for Moorchchhana generation) on a chosen position on a string of the Veena. In this case, as per Bharata’s description the Vadi swaraor AadharSwara or Shadjaswar ,is established on the fourth fret of the Bharat Veenaas shown in the above figure.

Establish Tara Shadjaswara by using Shadja-Tara Shadja Bhava or Sa-higher Sa consonance. This step is not very difficult because even a novice in music can easily notice this consonance. Sharangadeva in his Sangita Ratnakar has already clearly mentioned this.

व्यवहारे त्वसौ त्रेधा हृदि मन्द्रोऽभिधीयते| कण्ठे मध्यो मूर्ध्नि तारो द्विगुणश्चोत्तर:||[Sangita Ratnakar 1-3-7]

By using Shadja-Panchama Bhava i.e., thirteen-shruti distance establishes a Panchama swara on the string as shown in the above photograph. Expert musicians can do this very easily due to their precise tonal sensitivity.

By using Shadja-Madhyama Bhava i.e., nine-shruti distance establishes a Madhyamaswara on the string. All this is based on the Bharata’s description.

Then you get a Ma-Pa interval which is supposed to be of four shruti distance according to Bharata’s description.

Now it is possible to establish Bharata’s Gandhara on this basis because the interval between Bharata’s Gandhara and Madhyama is of four shrutis according to Bharata.

From Bharata’s Gandhara, Bharata’s Nishada can be derived by applying Sa-Pa Bhava.  The next step is quite interesting.

It is possible to establish Chatuh Shruti Rishabha by following Ma-Pa Bhava because it is four shrutis away from Shadja. So as an intermediate step Chatuh Shruti Rishabha is generated though it is not a part of Bharata’s Saptaka. The next step is even more interesting.

Interval between Bharata’s Gandhara and Chatuh-Shruti Rishabha can be judged and experienced. It is one shruti interval. Using this interval, Bharata’s Rishabha can be generated as follows. Bharata’s Rishabha is lower than the Chatuh shruti Rishabha.

Thus Chatuh-Shruti Rishabha helps in inferring Bharata’s Rishabha because the interval between Bharata’s Rishabha and Chatuh-Shruti Rishabha is also of one shruti. As per the description there is a two shruti distance between Bharata’s Rishabha and Bharata’s Gandhara and Chatuh-Shruti Rishabha falls exactly at the midpoint between them.

[This can be done by another method as well. Once Bharata’s Gandhara is established as discussed earlier the two-shruti distance is distinct and easy to recognize which is a `Vivadi’ distance. So, using this distance one can derive Bharata’s Rishabha from Bharata’s Gandhara because as mentioned by Bharata the distance between Bharata’s Rishabha and Bharata’s Gandhara is of two shrutis.]

From Bharata’s Rishabha, by applying Shadja-Panchama Bhava, Bharata’s Dhaivata can be established as shown in the above image.

Thus, all the basic seven swaras of Bharata’s Saptaka can be established by using the above method. There are few other methods for establishing a Saptaka on Bharat Veena. However, these are not discussed here for the brevity of description.

Shruti Generation by using ChakriyaNyaya method

As already shown above by using Shadja-Panchama Bhava and Shadja-Madhyama Bhava as well as Vivadi Bhava, we can establish a Saptaka of seven swaras in a step-by-step manner. Similarly, by using Chakriya Nyaya method indicated in Natyashastra and elaborately discussed in Abhinav Bharati by Abhinav Gupta, it is possible to establish all the twenty-two shrutis on the Bharata Veena. This is shown in following paragraphs in a step-by-step manner again. According to Natyashastra, Sa-Ma ratio is known as Shadja-Madhyama Bhava and the Sa-Pa ratio is known as Shadja-Panchama Bhava. The Chakriya Nyaya method is based on these two rules from Bharata’s Natyashastra. The Chakriya Nyaya method is also called `Swara-Mandala Sadhanam’.

अथद्वौ ग्रामौ षड्जग्रामो मध्यमग्रामश्चेति| अत्राश्रिता द्वाविंशतिश्रुतयः स्वरमण्डलसाधिताः|[Bharata’s Natyashastra]

तत एव स्थानान्तरे स्वरमंडलत्वमिति चक्रमुच्यते| तच्च परिमंडलं आंगिरसकाश्यपादिभिः मुनिभिः दर्शितम् |  [Abhinavagupta in Bharata, p. 19].

This method was known to the ancient sages such as Aangirasa and Kashyapa.  Abhinava Gupta has described this method in his commentary on Bharata’s Natyashastra. The process goes on as shown in following images.

You start the process with the Shadjaswara or Aadharswara. Apply the Shadja-MadhyamaBhava and Shadja-Panchama Bhava to Sa swara and generate Ma swara and Pa swara respectively. Then again apply both the Bhavas respectively to newly generated Madhyama and Panchama swara to get Bharata’s Nishad and higher Chatuh shrutiRishabha as shown in the above image. This higher Chatuh shrutiRishabha falls outside the bounds of a Saptakaso we need to bring it within the bounds of Saptaka by applying Sa-higher Sa Bhava in a reverse way. This whole process can be continued in a cyclical fashion till you get twenty-two shrutis. If you continue the process, you will get the same shruti places again and again. Therefore, this process is called ‘Chakriya Nyaya’ method as shown below.

`Shruti-Nidarshanam’ Experiment

After establishing shrutis, the Shruti-Nidarshanam experiment from Bharata’s Natyashastra can be performed on the Bharat Veena. Here only salient steps or ‘Saranas’ of the experiment are shown below. For the details of this experiment, you can refer a monograph titled ‘The Doctrine of Shrutis in Indian Music’, by the author. Four steps of the experiment are as follows.

First Sarana

Second Sarana

Third Sarana

Fourth Sarana

Shruti-Nidarshanam experiment demonstrates Samvad Bhavas or consonances among swaras such as Sa-Pa, Sa-Ma, BRe-BDha, Bga-Bni, Re-Dha, Ga, Ni and so on…This experiment provides empirical and experimental basis for Samvadi Swaras, Vivadi Swaras and Anuvadi Swaras. These Shruti, Swaras, Samvad-Bhavas between and among swaras lead to many more musical concepts such as Jatis, Moorchchhanas, and Taanas. Ancient music was different from contemporary music. However, these concepts eventually evolved into the system of Ragas. This discussion indicates that artistic expressions can be understood as computations. Underneath the aesthetic experience evoked by a piece of music or a musical composition there lies a hidden grammar of music. A creative person has a good understanding of the grammar so that he can explore end-less possibilities of the grammar to generate aesthetic experience. A creator is not only a creator of the work of art but a creator of the aesthetic experience.

Using computational algorithms artists can explore new structures, leading to new forms of cultural significance. ‘New Aesthetics’ for a ‘New Age’. Indian Arts inherit computational foundations as discussed above, therefore, in this ‘Digital Age’ there is an opportunity for artists to see new genres of art practices, new aesthetics, new forms of representations, and a whole new artistic creativity. Computational Music for instance…

Indian Artificial Intelligence Music.

The author of this paper has designed and developed an AI system called AIRaga based on the concepts and principles described in Natyashastra and Sangita Ratnakar. During the conference a demonstration of the system is given to show how the discussion so far in above paragraphs is relevant in the age of digital technology as well as how computational grammar of Indian music leads to generation of aesthetic experience.

AI Raga’ System

The main objective of ‘AI Raga’ is to show that AI can be used to simulate Ragas of Indian Music. In this case, AI enters into the realms of human creativity, by composing Gat/Bandish. The AI Raga system can generate Ragas and composes a Gat/Bandish and renders it on artificial instruments accompanied by artificial Tabla. This system has the encoded knowledge of all the concepts required for generating and playing a Raga. These concepts are Shruti, Vadi, Samvadi, Graha, Nyasa, Shadava, Oudava etc. and the knowledge of the principles such as the Shadja-Panchama Bhava, and the Shadja-Madhyama Bhava from the Natyashastra and Sangeet Ratnakar. AIRaga system was presented during the conference.

Through simulation and advanced computation, the system generates a musical composition in a Raga as given by the user. In the following image, the user has selected Raga Malakans. The user wants the system to generate a composition in Tritala in DrutaLaya that is fast tempo. The user has selected the e-Sarod as an instrument. This instrument is a digital instrument. This instrument does not exist in the real world. The sound of this instrument is created by using digital sound synthesis technology. Once the user clicks, the AI Raga system generates a composition in Raga Malakans.

In the following image you can see the Raga name as Malakans and all the relevant technical information such as the basic frequency of ‘Sa’ swara. Then it says Bharata’s Natyashastra related Shruti data is generated. The AI Raga system does not have fixed database of swaras or Ragas. Once the name of the Raga is given, the system generates the entire data specific to that Raga. The image shows that the Natyashastra related Graha-Nyasadiswaras are generated. The characteristic Meends, Aalaps, Ragangas and all required data is generated by the system and then a composition or Gat/Bandish is generated and that is played by the computer and the user can listen to it.

During the conference Raga Malakans was generated by the system and Gat/Bandish was played for the audience by the AI Raga system.

One can listen to the outputs of the AIRaga and AITala, another AI system to generate Talas at

https://www.youtube.com/channel/UCm_QHdRyO9kcmDqx_OY0KnQ and at

http://computationalmusic.com/latest_compositions.php

Through simulation of Ragas, concepts such as Shruti, Vadi, Samvadi, Graha, Nyasa, Shadava, Oudava etc. and principles such as…Shadja-Panchama Bhava, Shadja-Madhyama Bhava from Natyashastra are validated. Through simulation of Ragas, we can validate the science of Indian music. An attempt is made to validate principles of music mentioned in Natyashastra and Sangeeta Ratnakar. The main objective of this research is to show that digitization, computation & AI can be used to document, preserve, interpret and simulate the concepts from Indian Music, and take them forward creatively.

This is possible with almost all Indic knowledge systems due to the inherent logical structure present in the sutra treatises. By demonstrating the capabilities of AI and its use in creating Indian classical music, this paper seeks to assert that AI has tremendous potential to enable and supplement the study of Indic studies.

After the presentation, many relevant issues and questions were discussed. Some of the questions were about the AI technology and some other questions were about the justification of such an AI creation. Some questions were about the aesthetic experience in the context of AI. These issues and questions are briefly discussed here as follows.

Currently many researchers are working in the areas related to Artificial Intelligence (AI) and Indian music. A question came up that what is the difference between the technology that focuses on generating ‘Gamaka’ and this AI system? The answer is that this technology is completely different and secondly, AIRaga system generates a Bandish or a Raga composition without any human support or interference. Other technologies need human support or human intervention. The other question was about the Apps which use AI for accompanying the performer. The major difference is that the App that uses AI for accompanying a performer plays the role of an accompanist. Its role is to follow and imitate what the performer is doing in a best possible way. The AIRaga system is a performer itself. It generates a new composition every time and performs it. Therefore, it is completely different from the Apps that accompany a performer. Then there was also a question whether ‘Gamaka’ is possible with this system. The answer is ‘Yes’, though currently, it is not fully explored for all the instruments. The theory and algorithms for Gamaka generation are already in place. One very interesting question came up as ‘Can a Malashri Raga of Gandhar Gramabe played on the system?’ The answer to this question is that AIRaga system does not have a pre-defined database of Ragas. If you provide Aroha and Avroha of the Raga and Vadi and Samvadi swaras of the Raga then the system can do all the necessary computations and generate the Raga composition. This system does not have pre-defined database of Malashri Raga. So, if the Aroha and Avrohaas well the Vadi and Samvadi swaras of Malashri are provided to the AIRaga system it will generate the composition in Raga Malashri. One of the audiences felt that the Malakans that is generated by the system during the conference presentation, does not sound like a Malakans. This comment has multiple dimensions involving subjectivity. However, keeping aside subjective elements, one can say so. There is a reason behind it. We remember and recognize Ragas based on known compositions and tunes. Raga Malakans is a well-established Raga and there are many popular compositions available. We tend to compare the AI generated composition with existing compositions of Raga Malakans. In this case, however, when a system generates a composition, it does not refer to the existing compositions. It generates a fresh new composition based on rules. So, it is always possible that the composition generated by the system sounds different than known compositions. The system is now well-tested and does not deviate much from the existing established compositions. Some deviation is always necessary to bring in the element of novelty. Otherwise, new composition will not be generated. A similar question was also raised asking whether the system has been given any training inputs. Actually, this system does not require training. This technology is not based on ‘Artificial Neural Network’ (ANN). This technology is based on the principles of Natyashastra. Secondly, this technology is very similar to the ‘Optimality Theory’ that is used mostly in modern linguistics. So, it does not require any training data set. Then the next question follows that then how to validate the output of the system? Answer is that the system output is evaluated on the basis of whether it is following rules and principles or not. In this sense there is a ‘Shastric’ validation. The generated compositions are consistent with the principles and rules outlined in Natyashastra and therefore they are valid. Apart from this, the system is validated by giving demonstrations and having discussions with practicing musicians and music researchers. This is the way the AIRaga system is fine-tuned.

Another issue that naturally comes to the mind is ‘Can Artificial Intelligence (AI) create ‘Bhava’ in music?’ One of the views is that, when an artist is performing, he or she is possessed with particular emotions. However, AI does not have emotions or ‘Bhava’ by itself, then how AI can replicate ‘Bhava’ into AI music performance. Among other responses to this issue could be a rather radical response that the AI performance produces ‘artificial’ ‘Bhava’. In this case, ‘Bhavas’ are not encoded in the AI system on the other hand, the ‘Bhava’ is a resultant experience to the audience. This needs some brief explanation, which follows in the next few paragraphs.

The AI perspective would be that one needs to look at aesthetic experience by keeping the ‘creation’ in the focus where the ‘creator’ of a work of art does not have much role to play. For instance, the beautiful sunrise or beautiful sunset evokes an aesthetic experience. A beautiful flower evokes an aesthetic experience. The capability of evoking an aesthetic experience in the minds of the perceiver or the audience remains in the characteristic structure and features of the ‘aesthetic object’. A significant set of characteristic features of the object or the work of art is responsible for the evocation of aesthetic experience in the minds of the observer or the audience. The sunrise and the sunset evoke an emotional appeal, may be due to specific colors and their schematization and the twilight settings of transitions from night to day and day to night respectively. These characteristics contribute to evoke emotions, feelings, aesthetic experience and even psychological impact. In the case of AI creations too, it is very difficult to assign a status of a creator to AI, equivalent to human creator. Therefore, in the case of AI creations, if the creations have significant structure and capability of evoking the aesthetic emotions or ‘Bhava’ then it really does not matter whether the creation is created by an AI or it is created by humans.  AIRaga system generated compositions have artistic elements organized in it by following the rules of the grammar of Indian music in such a way that they are able to evoke aesthetic experience in the minds of the audience.

The next argument in favor of AI creations would be to say that currently, AI music is like an instrumental music. For instance, instrumental music does not have words. There are only abstract patterns of swaras structured in a significant way so that they are capable of evoking aesthetic emotions. Here the grammar of music plays an important role in achieving this. Therefore, if AI also takes up the same approach and through appropriate application of musical grammar generates a musical composition and successfully evokes the aesthetic experience then it should be treated valid aesthetic experience. There are many forms of musical presentations where there are no lyrics used but meaningless phrases using some syllables are used in the compositions. For instance, a popular musical form called `Tarana’ in Hindustani music does not use any meaningful words or lyrics in the composition. There are no meaningful words, still it is a very popular art form which is enjoyed by the audience. There are beautiful patterns of musical notes.

With these arguments it can be said that the arguments raised against the capability of AI creations with regards to evocation of aesthetic experience are not sound. It is not necessary to always equate the aesthetic experience evoked by AI creations with the aesthetic experience evoked by works of art created by humans. This comparison is not tenable beyond a point.

During the discussion many related points and other allied issues were discussed. Following weblinks may be referred to go into more details of these issues.

The Doctrine of Shruti in Indian Music’ by the author of this paper at:

http://computationalmusic.com/the-doctrine-of-shruti-in-indian-music.php

Shruti Nidarshanam or Sarana ChatushtayiPrayog:

http://computationalmusic.com/shruti-nidarshanam-the-sarana-chatushtaya-experiment.php

Swara-Sthapana on Bharat Veena:

http://computationalmusic.com/swara-sthapana-on-shruti-veena-or-shruti-sitar.php

Bharat Veena:

http://computationalmusic.com/bharat-veena.php

You can listen to Indian AI Music at-

https://www.youtube.com/channel/UCm_QHdRyO9kcmDqx_OY0KnQ

You can visit the website for more information at

http://computationalmusic.com/

Acknowledgements: I would like to thank my wife Archana Vidwans and my son Vihang Vidwans for helping me in preparing the final draft of this paper.

References:

  1. ‘Agnipuranam’ (Sanskrit), Acharya Shivaprasad Dwivedi (Hindi Comm.), Choukhamba Sanskrit Pratishathan, New Delhi, 2004, Ch. 339.10
  2. Anandavardhana, ‘Dhwanyaloka’, Dr. Nagendra (Ed.) Acharya VishweshwaraSiddhanta Shiromani (Hindi Comm.), Dnyana Mandal Ltd, Varanashi, 1962, p. 312
  3. Apte, Vaman S. (Ed.) (1970), The Student’s Sanskrit-English Dictionary, Published by Motilal Banarasidas, Delhi, 2nd revision 1970, 6th reprint 1988
  4. Bharata (Date not known), Natyashastra of Bharatamuni, with the commentary- Abhinavabharati by AbhinavaGuptacharya (Chapters 28-37), Vol. IV, edited by Kavi, M. Ramakrishna and Pade, J. S., Baroda: Oriental Institute Baroda, 1964.
  5. Bhise, Usha R. (Ed.), (1986), Naradiya Shiksha, Original Commentary by Bhatta Shobhakara, critically edited with translation and explanatory notes in English by Dr. Usha R. Bhise, Poona: Bhandarkar Oriental Research Institute, Poona, India.
  6. Brihaspati, Acharya (1986), Natyashastra, 28th Chapter –Swaradhyaya, (Hindi) with Sanskrit and Hindi commentary, New Delhi: Brihaspati Publication.
  7. Jog, D, V. (Ed), Subodh Upanishatsamgraha (Marathi), Part 1, Taittiriyopanishad- Brahmavalli, 2018, pp. 221-222
  8. Kuanpoonpol, Priyavat, ‘Pratibha: The Concept of Intuition in the Philosophy of Abhinavagupta’, Harvard University, Cambridge, Massachusetts, 1993
  9. Osborne Harold (Ed.),The Oxford Companion to Art, Oxford University Press, London, 1970, 1975, p.-12
  10. Rajasekhara, ‘Kavyamimansa’, Dalal, C. D., and Shastri, R. A., Rama Swami Shastri Shiromani, K. S., (Eds.) ‘Kavyamimansa of Rajasekhara’, Oriental Institute, Baroda, 1034, Chapter IV, p.-15
  11. Sharangadeva, Nishshanka (13th AD), Sangita Ratnakar (original Sanskrit text), translated by Taralekar, G. H. (Marathi Trans.), Mumbai: Maharashtra Rajya Sahitya Sanskriti Mandala, Published in 1975
  12. Urmson, J. O. and Ree J. (Ed.) in ‘The Concise Cyclopaedia of Western Philosophy and Philosophers’, Pub. Unwin Hyman Inc., London, UK, 1989, p.-2

Conference on Hindu Aesthetics

Watch video presentation of the above paper here:


Disclaimer: The opinions expressed in this article belong to the author. Indic Today is neither responsible nor liable for the accuracy, completeness, suitability, or validity of any information in the article.

Leave a Reply

More Articles By Author