Pedro Ramoneda

Bringing together Music Education and AI

About

Technology and music are my passions. It is my primary motivation to pursue research in sound and music computing and music information retrieval. I have previously researched on Harmony-related topics, and I am currently pursuing my PhD studies on the intersection between music performance and artificial intelligence at Music Technology Group , Universitat Pompeu Fabra, Barcelona.

My PhD thesis, "Can I play it? Understanding piano performance difficulty through explainable and multimodal machine learning" focuses on audio and symbolic music dataset creation and deep learning for performance difficulty analysis under the supervision of Prof. Xavier Serra . We aim to understand the performance difficulty, through pedagogically motivated representations, to recommend a customised and optimised learning path for each music student and allow musicians explore large music score collections. My ML research interests include explainable machine learning, generative models, human-centered AI and curriculum learning.

During the summer of 2023, I embarked on a research internship under the supervision of Taketo Akama at Sony CSL (Tokyo). My attention was captured by the fascinating field of music generation, particularly the challenge of inpainting to complete missing sections of musical compositions. This hands-on experience was crucial in deepening my understanding of music generation, and it has significantly heightened my interest in this area of study. Currently, I am actively collaborating with my colleague and mentor, Dasaem Jeong, on developing innovative music generation systems that are specifically designed to enhance music education. Together, we are working to create user-friendly and effective tools that support both educators and learners in their musical journeys.

Papers

Here you can find the list of my publications, as well as the resources associated with them (when available). You can also check my list of publications on Google Scholar.

P. Ramoneda, E. Parada-Cabaleiro, B. Weck, X. Serra (2024). The Role of Large Language Models in Musicology: Are We Ready to Trust the Machines? . In Proc. of the 3rd Workshop on NLP for Music and Audio (NLP4MuSA), Satellite of ISMIR, 2024.

Paper Benchmark

P. Ramoneda, M. Rocamora, T. Akama (2024). Music Proofreading with RefinPaint: Where and How to Modify Compositions given Context . In Proc. of the 25th Int. Society for Music Information Retrieval Conf. (ISMIR), San Francisco, 2024.

Paper Demo Code

P. Ramoneda, V. Eremenko, A. D'Hooge, E. Parada-Cabaleiro, X. Serra (2024). Towards Explainable and Interpretable Musical Difficulty Estimation: A Parameter-efficient Approach . In Proc. of the 25th Int. Society for Music Information Retrieval Conf. (ISMIR), San Francisco, USA..

Paper Demo Code

P. Ramoneda, M. Lee, D. Jeong, J. J. Valero-Mas, X. Serra (2024). Can Audio Reveal Music Performance Difficulty? Insights from the Piano Syllabus Dataset . Submitted to TASLP.

Paper

P. Ramoneda, J. J. Valero-Mas, D. Jeong and X. Serra (2023). Predicting performance difficulty from piano sheet music images . In Proc. of the 24th Int. Society for Music Information Retrieval Conf., Milan, Italy..

Code Paper Dataset Demo

Ramoneda, P., Jeong, D., Eremenko, V., Tamer, N. C., Miron, M. and Serra, X. (2024). Combining piano performance dimensions for score difficulty classification . Expert Systems with Applications, 238.

Code Paper Dataset Demo

N. C. Tamer, P. Ramoneda, and X. Serra (2022). Violin Etudes: A comprehensive Dataset for f0 Estimation and Performance Analysis . In Proc. of the 23rd Int. Society for Music Information Retrieval Conf. (ISMIR).

Code Paper Dataset Teaser

Pedro Ramoneda, Dasaem Jeong, Eita Nakamura, Xavier Serra, and Marius Miron (2022). Automatic Piano Fingering from Partially Annotated Scores using Autoregressive Neural Networks . In Proceedings of the 30th ACM International Conference on Multimedia (MM ’22), October 10–14, 2022, Lisboa, Portugal..

Code Paper Dataset Demo Teaser

Ramoneda, P., Tamer, N. C., Eremenko, V., Miron, M. & Serra, X (2022). Score difficulty analysis for piano performance education based on fingering . In ICASSP 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

BibTeX Arxiv Code Dataset Demo Teaser

Kim, H., Ramoneda, P., V., Miron, M. & Serra, X (2022). An overview of automatic piano performance assessment within the music education context . In Proceedings of the 14th International Conference on Computer Supported Education (CSEDU).

BibTeX PDF

Bernardes, G., Ramoneda, P. and Miron, M (2021). Unveiling High-level Discriminant Harmonic Descriptors of Musical Style in the Tonal Interval Space . In International Conference on Music Perception and Cognition (ICMPC16-ESCOM11).

BibTeX Pre-print Supplementary Material

Ramoneda, P., Miron, M and Serra, X (2021). Piano Fingering with Reinforcement Learning . In CoRR.

BibTeX Pre-print Code

Ramoneda, P. and Bernardes, G. (2020). Revisiting Hamonic Change Detection . In 149th AES convention, the Audio Engineering Society.

BibTeX Pre-print Code

Dissertations

Master’s thesis, Universitat Pompeu Fabra, Barcelona. Title: Computational methods to study piano music in education context. Under the supevision of Dr. Marius Miron and Prof. Xavier Serra.

Bachelor’s thesis, University of Porto and University of Zaragoza. Title: Harmonic change detection from musical audio. Under the supevision of Prof. Gilberto Bernardes.

Teaching

I have had the privilege of serving as a Teaching Assistant to Prof. Xavier Serra, contributing to various courses and practical sessions. My responsibilities and contributions include:

  • Music Technology Lab (2022): Hands-on experience and practical knowledge in music technology and software development (17 hours).
  • Signals and Systems (2022): Conducted labs and exercises, helping students grasp complex concepts in signals and systems (42 hours).
  • Music Technology Lab (2023): Lectures to enrich students’ understanding of music technology and software development (17 hours).
  • Signals and Systems (2023): Conducted labs and exercises, ensuring students’ proficiency in signals and systems (42 hours).
  • Music Technology Lab (2024): Lectures to enrich students’ understanding of music technology and software development (17 hours).
  • Signals and Systems (2024): Conducted labs and exercises, ensuring students’ proficiency in signals and systems (42 hours).

Supervision

  • Music Corpus Annotator: A Local Annotation Tool for Symbolic Sheet Music in Music Information Research - Bachelor thesis by Nerea Sastre (2023). (Maximum grade: 10/10)
  • Automatic Score-to-Score Music Generation - Master thesis by Quoc Duong Nguyen (2023). Co-supervision with Carlos Hernandez Olivan.
  • Piano Performance Analysis Using Technique Information Extracted from Videos - Master thesis by Samuel Cantor (2023). Supervising research focused on extracting and analyzing piano performance techniques using video data. Co-supervision with Nazif Can Tamer.
  • Computational Segmentation of Sheet Music Books for Search and Retrieval - Master thesis by Samuel Gómez Castro (2024). Supervising research focused on Music sheet Images segmentation.
  • Esmuc Interns - Pedro D'Avila (2023) and Lucía Catalán (2024).

Talks

  • Seminar at COLT UPF Linguistics Department (November 2024): "Bridging NLP and Music: From Explainable Difficulty Analysis to LLMs in Musicology". Speaker: Pedro Ramoneda.
  • Talk at Open Data Week (October 2024) at Universitat Pompeu Fabra: "Evaluación de la Interpretabilidad: Análisis de la Dificultad de Ejecución en Piano mediante Aprendizaje Automático Explicable y Multimodal". Speaker: Pedro Ramoneda.
  • Conservatorio de Alcañiz (April 2024): "Music Information Research: Analizando Interpretaciones Musicales Automáticamente". Speaker: Pedro Ramoneda.
  • Lecture at UNIR (March 2024): "Music Information Retrieval and Music Education, Supporting Not Replacing". Speaker: Pedro Ramoneda.
  • Seminar at University of Alicante (February 2024): "Analyzing Piano Performance Difficulty with Explainable Multimodal Machine Learning". Speaker: Pedro Ramoneda.
  • DLBCN Deep Learning Barcelona Symposium (2023): "Predicting Performance Difficulty from Multimodality". Speaker: Pedro Ramoneda (UPF).
  • DLBCN Deep Learning Barcelona Symposium (2022): "Integrating Autoregressive Neural Networks for Comprehensive Piano Fingering and Score Difficulty Analysis". Speaker: Pedro Ramoneda.
  • Lecture (2022) at ESMUC: "Why Music Information Research (MIR)?" - A presentation to 4th-year students of the musicology bachelor program, invited by Luca Chiantore. Speaker: Pedro Ramoneda.
  • Seminar (September 2021): "Reproducibility in Signal Processing Research" - Mirdata, dataset loaders for reproducible research on Music Information Retrieval. Speakers: Genis Plaja, Pedro Ramoneda.
  • Talk (July 2021) at ICMPC-ESCOM: "Unveiling High-level Discriminant Harmonic Features of Musical Style in the Tonal Interval Space". Speaker: Pedro Ramoneda.

Research Visits

  • Universidade do Porto (2019-2020): Full academic year as an Erasmus scholar, collaborating with Gilberto Bernardes on automatic harmonic analysis.
  • Sony CSL (June 2023 - September 2023): Funded by Sony CSL, collaborating with Taketo Akama on automatic music generation.
  • University of Alicante (February 2024): Funded by a the University of Alicante grant, one-month research visit with José Javier Valero-Mas, focusing on audio classification and Optical Music Recognition (OMR).

Awards

  • Open Science Award (2024): Awarded by Universitat Pompeu Fabra for the best use of open data in a doctoral thesis.

Software Projects

I am passionate about developing innovative software solutions that contribute to the field of Music Information Research. My projects encompass a wide range of applications, from data analysis tools to interactive applications to enhance the user’s music education experience.

You can explore my extensive portfolio of software projects on my GitHub page. Here, you will find the source code, documentation, and detailed explanations of each project, providing insights into my development process and the technologies I employ.

In addition to my open-source projects, I have developed demos for industry outreach to showcase the practical applications of my research projects. These demos are designed to provide a tangible experience of the innovations and solutions derived from my work. You can interact with these demos on the MusicCritic platform.