Genís Plaja-Roglans


I am a research scientist at Moises. Previously, I completed my PhD at the Music Technology Group at Universitat Pompeu Fabra under the supervision of Prof. Xavier Serra, Prof. Martín Rocamora, and Dr. Marius Miron. My research interests are–but not limited to–music source separation, deep generative modeling, melody estimation. During the PhD I also worked on Carnatic music, exploring repertoire-specific techinques to adapt models to this art music tradition. I am currently based in Barcelona, visiting my childhood village in l'Empordà whenever I can.

Projects I contributed to

compIAM

Standardized and MIR-friendly implementations to tools and access to datasets for the computational analysis of Indian Art Music

dmx-diffusion

Modular Python package for music source separation experiments using diffusion models.

Education and work experience

Education

  • PhD student at Music Technology Group

    (Sept. 2021 - Feb. 2026) // Universitat Pompeu Fabra
    Under the supervision of Prof. Xavier Serra
    *Completed with grade of Excellent and cum laude mention.

  • Master in Sound and Music Computing

    (Sept. 2020 - Sept. 2021) // Universitat Pompeu Fabra

  • International exchange program in Australia

    Feb. 2020 - July 2020) // Royal Melbourne Institute of Tech.

  • Undergraduate Degree in Audiovisual Systems Engineering

    (Sept. 2016 - July 2020) // Universitat Pompeu Fabra

Work experience

  • Data Scientist at Moises AI

    (Aug. 2025 - ongoing) // Moises AI

  • Research internship at Moises AI

    (Apr. 2023 - Dec. 2023) // Moises AI
    Working on music source separation using generative modeling under the supervision of Dr. Igor Gadelha.

  • Acousting Engineering teacher

    (Sept. - Dec. 2021-2024) // Universitat Pompeu Fabra
    Given to the 3rd year of the Bachelor's Degree in Audiovisual Systems Engineering.

  • Research assistant (MTG)

    (Sept. 2020 - Aug. 2021) // Universitat Pompeu Fabra
    Contributing to the mirdata and soundata libraries by writing dataset loaders for datasets in the MTG, also contributing to the core development and maintenance of both libraries.

  • Research intern (MTG)

    (Nov. 2018 - Jan. 2020) // Universitat Pompeu Fabra
    Leading the Sounds of Science project, an initiative to promote research through the matter of sound. We also collected sounds from several different science departments, created a database, and performed labelling.

Selected publications

→ G. Plaja-Roglans, X. Serra & M. Rocamora, 2025. Leveraging Carnatic live recordings for singing voice separation using regression-guided Latent Diffusion. In: Proc. of the 25th Int. Society for Music Information Retrieval Conf. (ISMIR), Daejeon, Korea. Link to paper.

→ G. Plaja-Roglans, Y. Hung, X. Serra & I. Pereira, 2025. Efficient and fast generative-based singing voice separation using a Latent Diffusion Model. In: Proc. of the Int. Joint Conference of Neural Networks (IJCNN), Rome, Italy. Link to paper.

→ G. Plaja-Roglans, Y. Hung, X. Serra & I. Pereira, 2025. Generating separated singing vocals using a diffusion model conditioned on music mixtures. In: Proc. of the IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA), Lake Tahoe, USA. Link to paper.

→ G. Plaja-Roglans, M. Miron, A. Shankar, X. Serra, 2023. Carnatic singing voice separation using cold diffusion on training data with bleeding. In: Proc. of the 24rd Int. Society for Music Information Retrieval Conference (ISMIR 2023), Milan, Italy. Link to paper.

→ G. Plaja-Roglans, M. Miron and X. Serra, 2022. A diffusion-inspired training strategy for singing voice extraction in the waveform domain. In: Proc. of the 23rd Int. Society for Music Information Retrieval Conference (ISMIR 2022), Bengaluru, India. Link to paper.

→ M. Fuentes, G. Plaja-Roglans, G. Cortès-Sebastà, T. Khandelwal, M. Miron, X. Serra, J. P. Bello, and J. Salamon, 2024. Soundata: Reproducible use of audio datasets. Journal of Open Source Software, 9(98), 6634. Link to paper.

→ T. Nuttall, G. Plaja-Roglans, L. Pearson and X. Serra, 2022. In search of sañcaras: tradition-informed repeated melodic pattern recognition in Carnatic Music. In: Proc. of the 23rd Int. Society for Music Information Retrieval Conference (ISMIR 2022), Bengaluru, India. Link to paper.

→ T. Nuttall,G. Plaja-Roglans, L. Pearson, X. Serra, 2021. “The Matrix Profile for Automated Discovery of Repeated Motifs in Audio - Indian Carnatic Music”. In: International Symposium on Computer Music Multidisciplinary Research (CMMR), Tokyo, Japan. Link to paper.


Additional recent research contributions

→ Teaching assistant at the 2nd Generative AI Workshop.

→ "mirdata: dataset loaders for reproducible research in MIR" in the Seminar for Reproducible Research organized by GdR ISIS (France).

→ Lecturer and main developer in the ISMIR 2022 Tutorial on Computational Methods for Supporting Corpus-Based Research on Indian Art Music.

→ Speaker and organizer in CompMusic workshop, organized in Chennai, India, as a satellite event of ISMIR 2022.