Scope

India is replete with diverse and thriving art, folk and film music cultures. These music cultures are backed by centuries of evolution in performance, practice, music instruments, pedagogy, musicology and a dedicated community. As one of the fastest growing emerging markets, India provides multiple opportunities for the growth of music and music applications, necessitating the research and development of technologies for improved production and consumption of music. The unique combination of sophisticated and diverse music cultures, coupled with growing applications has renewed interest in a multitude of research problems in Music Information Research (MIR) for Indian music, while also posing new challenges and providing novel opportunities.

MIR spans a wide range of problems and tools grounded in signal processing and machine learning, and has had a significant presence within ICASSP over the last two decades. However, there have been challenges to building MIR methods for analysis and synthesis of Indian music. Some of the unique aspects of Indian music are the dominance and sophistication of oral traditions, under-explored instruments and performance styles, diverse music cultures, and limited adoption of technology. Consequently, there has been a scarcity of annotated datasets and meaningful evaluation metrics linked to specific problems rooted in Indian music. Further, the growing global attention to music generation research raises interest in the potential of domain-specific representations.

While recent advances in machine learning have proliferated across various areas such as text, handwriting, speech, vision, biological signals, music and communication, there are very few efforts where domain knowledge is effectively applied. With music being so closely linked to particular societies and the culture, it is but natural that to meaningfully address the most relevant and interesting problems demands an understanding of all these aspects. Past work from CompMusic project www.compmusic.upf.edu strongly suggests that “a knowledge based signal processing approach”, combined with machine learning, can boost the effectiveness of general music processing methods on Indian art music significantly. Hence, novel emerging applications with culture-aware methods that can effectively integrate socio-cultural aspects into research methods provide new opportunities for MIR in Indian music.

The objective of the workshop is to focus on analysis, generation, synthesis and applications for Indian music. Aiming to address the diverse current needs, challenges and opportunities in Indian music the workshop will focus on papers related to computational approaches to the analysis, modeling and generation of Indian music taking into account its multilingual, multicultural and regional nuances and diversity. The workshop will invite papers and foster discussions across interdisciplinary areas spanning audio and music signal processing, acoustics, machine learning, music cognition, musicology, music production and music education.

Specifically, the topics of the workshop, with its focus on Indian music, will cover (without being limited to): Standard MIR tasks for analysis of music features, music generation and synthesis, annotated datasets, music instrument modeling and analysis, musical acoustics, digital instruments, analysis and synthesis in low resource conditions, signal processing or machine learning approaches to analysis of specific features of diverse genres of Indian music, multimodal applications (combining video, text and audio), music applications in multilingual and multicultural societies (such as dubbing), human-machine co-creation, music production, consumption and learning, applications to music heritage.