I participated at the Doctoral Students Work organized by my University. I presented a poster about my recent work on how to generate data to train neural networks for a more robust source separation, in the case of classical music. This poster is based on the paper we submitted to the SMC conference and the code, data and models I published in github and zenodo. It seems that the University (through the library) hosts a repository, so I will backup some stuff there.
M. Miron, J. Janer, and E. Gomez, “Generating data to train convolutional neural networks for low latency classical music source separation” Sound and Music Computing Conference 2017 (submitted)
I shared the open science award with another researcher from the department, Monica Dominguez, who works on natural language processing.
Although I have less time now, with the deadline to deliver the thesis approaching, I will make sure code and data are accessible and explained well, so other people can use it.
research, news, source separation, open science, datasets, award