Talks/Teaching

TALKS/TEACHING

[My website has moved to https://serrjoa.github.io/]

Talks & Tutorials

Imagination in power! Deep generative models with a focus on speech and music: Casa de Cultura de Girona (12/5/2022).

Universal, generative speech enhancement: Keynote at the Deep Learning Barcelona Symp. 2021 (23/12/2021) [YouTube, min 3].

Deep generative audio at Dolby: Universitat de Girona - Cicle de conferències Patronat/EPS (30/11/2021).

Understanding and visualizing speech quality: Keynote at the Web Audio Conf. 2021 (13/7/2021) [YouTube]. With J. DeLancey.

Version identification in the 20s: Tutorial at the Int. Soc. for Music Information Retrieval Conf. (11/10/2020) [ISMIR] [Slides]. With F. Yesiler & C. Tralie.

From correlation to imagination - GANs and deep generative models: Telefónica Talks - Madrid (27/9/2018), Telefónica Barcelona Talk (11/10/2018), Enginyeria La Salle - Universitat Ramon Llull (26/3/2019), Universitat de Vic - Jornades de Mecatrònica (9/5/2019), Universitat de Lleida - Màster Gestió Administrativa (27/5/2019), Conf. of the Catalan Association for Artificial Intelligence - CCIA (24/10/2019).

Deep learning at Telefónica Research: Dolby Labs - Barcelona (14/3/2018), Barcelona Technology School (26/6/2018).

Overcoming catastrophic forgetting with hard attention to the task: CVC-UAB - Lifelong Learning Seminar (16/2/2018), bcn.ai (6/9/2018).

Unintuitive properties of deep neural networks: UPC/TelecomBCN - IDL Winter School (30/1/2018), EU Science Hub - Human Behavior and Machine Intelligence Workshop (5/3/2018).

Facts and myths about deep learning: DataBeersBCN (4/10/2016) [YouTube], UPF MTG-Compmusic Seminar (18/11/2016) [YouTube], UPC-TelecomBCN DLSL Winter School (27/1/2017) [YouTube], B-Debate on Artificial Intelligence (7-8/3/2017) [Vimeo], IIIA-CSIC Seminars (15/3/2017), Madrid Machine Learning (11/5/2017).

Note onset deviations as musical piece signatures: UPF - Música i Publicitat (12/11/2014).

Patterns, regularities, and evolution of contemporary western popular music: CRM - CAMP seminar (3/5/2012), UPF - Música i Publicitat (6/11/2012, 13/11/2013).

Machine learning for music discovery: Tutorial at the Int. Symp. on Frontiers of Research on Speech and Music (20/01/2012) [CompMusic] [FRSM] [Handouts].

Audio content-based music retrieval: Tutorial at the Int. Soc. for Music Information Retrieval Conf. (24/10/2011) [MTG] [ISMIR] [Handouts]. With M. Müller.

Model-based cover song detection via threshold autoregressive forecasts: UPC - TALP (6/10/2010).

Music descriptor time series: Max Planck Institute for the Physics of Complex Systems (5/2/2010).

Audio cover song identification: UPF - Dept. of Information and Communication Technologies (25/10/2007).

Tecnologia per al descobriment de cançons i per la predicció d'èxits musicals: ESMUC - Seminaris de Sonologia (9/10/2006).

Teaching

[Currently I'm not teaching]

Invited lecturer, Universitat de Vic, Vic, Barcelona (2020). Postgraduate course on Artificial Intelligence with Deep Learning (2019-2020), Universitat Politècnica de Catalunya, Barcelona. Master in Sound and Music Computing (2019), Universitat Pompeu Fabra, Barcelona --- Deep learning seminars, theory, applications, & coding.

Part-time adjunct professor, Universitat de Vic, Vic, Barcelona. Undergraduate studies of Biomedical Engineering. Diagnosis Decision Support Systems: Deep learning seminars & coding (2018-2019).

Tenure-track lecturer accreditation ("Professor Lector") from the Catalan Government (Agency for Management of University and Research Grants, AGAUR), Jan 2014.

Teaching assistant, Dept. of Inf. and Com. Tech., Universitat Pompeu Fabra, Barcelona. Undergraduate studies. Probabilitat i processos estocàstics (2010-2011), càlcul mètodes numèrics (2009-2010), fonaments físics de la informàtica (2008-2010), computadors III (2006-2009).

Invited instructor, Eng. La Salle, Universitat Ramon Llull, Barcelona. MSc studies. Máster de Producción Sonora y Audio Digital. Music information retrieval (2007-2008).

Students

R.O. Araz. Improving quality and speed of universal speech enhancement with diffusion models. Student internship, Dolby Laboratories. Ongoing.

F. Yesiler. Data-driven musical version identification: accuracy, scalability, and bias perspectives. PhD thesis, Universitat Pompeu Fabra. 2018-Ongoing. Co-directed with E. Gómez.

G. Cambara. The effect of regressive and discriminative workers for learning self-supervised speech representations. Student internship, Dolby Laboratories. Sep 2021. Co-directed with S. Pascual.

J. Bustos. Towards audio-conditioned generation of music album artwork. MSc thesis, Universitat Pompeu Fabra. Sep 2020. Co-directed with P. Herrera. [DOI]

C. Steinmetz. Learning to mix with neural audio effects in the waveform domain. MSc thesis, Universitat Pompeu Fabra, & Student internship, Dolby Laboratories. Sep 2020. Co-directed with F. Font. [DOI]

S. Pascual. End-to-end speech synthesis using deep neural networks. PhD thesis, Universitat Politècnica de Catalunya. 2017-2020. Co-directed with A. Bonafonte.

M. Serra. Tunable, flow-based recommendations. Student internship, Telefónica Research. Dec 2019. Co-directed with C. Segura.

D. Álvarez. Out-of-distribution likelihoods in deep generative models. Student internship, Telefónica Research. Nov 2019.

M. Carós. Deep anomaly detection in machine-to-machine network logs. Student internship, Telefónica Research. Jul 2019. Co-directed with A. Lutu & D. Perino.

A. Gilbert. An investigation of in-sample forgetting in deep convolutional neural networks. MSc thesis, Universitat Pompeu Fabra. Jul 2019. Co-directed with M. Farrús.

J.F. Núñez. Normalizing flows for novelty detection. MSc thesis, Universitat Pompeu Fabra. Jul 2019. Co-directed with V. Gómez.

O. Slizovskaya. A prospective study of time series anomaly detection with normalizing flows. Student internship, Telefónica Research. Dec 2018. Co-directed with I. Leontiadis.

M. del Tredici. Graph convolutional networks for recommender systems. Student internship, Telefónica Research. Oct 2018. Co-directed with A. Karatzoglou, J. Luque, & C. Segura.

S. Raponi. User anonymity and reidentification in mobile weblog traces. Student internship, Telefónica Research. Aug 2018. Co-directed with N. Kourtellis, I. Leontiadis, & D. Perino.

J. Pons. Neural network architectures for few-instance audio classification. Student internship, Telefónica Research. Jul 2018.

D. Surís & M. Miron. Overcoming catastrophic forgetting in neural networks. Student internships, Telefónica Research. Dec 2017.

A. Pyrgelis. Assessing mobility trajectory uniqueness in telco networks. Student internship, Telefónica Research. Nov 2017. Co-directed with I. Leontiadis, N. Kourtellis, & C. Soriente.

S. Pascual. Deep spatiotemporal mobility prediction from cellular network events. Student internship, Telefónica Research. Dec 2016.

M.A. Orakzai. Gait-based authentication in real life. MSc thesis, Universitat Politècnica de Catalunya. Student internship, Telefónica Research. Jul 2016. Co-directed with A. Matic, L. Navarro, & C. Soriente.

G. Pelino & Z. Holler. Predicting the success of telemarketing campaign calls. MSc thesis, Barcelona Graduate School of Economics. Student internships, Telefónica Research. Jul 2016. Co-directed with A. Karatzoglou & A. Matic.

J.C. Vásquez-Correa. Analyzing the robustness of speech-based Parkinson detection algorithms under noise and audio transformations. Student internship, Telefónica Research. Feb 2016.

A. Bogomolov. Using call detail records for understanding crime data. Student internship, Telefónica Research. Jan 2016. Co-directed with N. Oliver.

F. Capó. Western classical composer identification using symbolic data. MSc thesis, Universitat Pompeu Fabra. Sep 2015. Co-directed with P. Herrera. [MTG]

F. Font. Tag recommendation using folksonomy information for online sound sharing platforms. PhD thesis, Universitat Pompeu Fabra. Jun 2015. Co-directed with X. Serra. [MTG]

G. Herrero. Towards supervised music structure annotation: a case-based fusion approach. MSc thesis, Universitat Pompeu Fabra. Sep 2014. [MTG]

M. Carbonell. Power laws: from linguistics to music. Bachelor's final project, Universitat Autònoma de Barcelona. Feb 2014. Co-directed with A. Corral. [CRM]

G. Meseguer. Automatic content-based detection of influences in the history of progressive rock music. MSc thesis, Universitat Pompeu Fabra. Sep 2013. Co-directed with P. Herrera. [MTG]

J. Van Balen. Automatic recognition of samples in musical audio. MSc thesis, Universitat Pompeu Fabra. Sep 2011. Co-directed with M. Haro. [MTG]

C. A. De Los Santos. Nonlinear audio recurrence analysis with application to music genre classification. MSc thesis, Universitat Pompeu Fabra. Sep 2010. Co-directed with R. G. Andrzejak. [MTG]

S. Bromberg. Recurrence quantification analysis in music information retrieval tasks: an example on genre classification. MSc thesis, Universitat Pompeu Fabra. Jun 2010. Co-directed with R. G. Andrzejak.

A. Almarza. Implementació i avaluació d’algorismes per la descripció i classificació rítmica de la música. Bachelor's final project, Universitat de Girona. Apr 2010. [alexalmarza]

C. Quirante. Quality assessment and enhancement of an industrial-strength audio fingerprinting system. MSc thesis, Universitat Pompeu Fabra. Sep 2009. Co-directed with P. Cano. [MTG]