We propose the PHENICX-Anechoic dataset which relies on the Aalto orchestral anechoic recordings. We denoised the original recordings and annotated each of the tracks corresponding to the instrument groups.
The Bach10 recordings synthesized with Sibelius can be found on the zenodo repository.
The note refinement code in part three of the thesis is on github. We do not have the rights to distribute the NMF framework and to some extent note refinement is integrated with this framework. However, multi-channel score-informed source separation can be computed using the Repovizz website. You need to create a datapack with datapack designer, upload it and select the checkbox which computes the source separation. The uploaded datapack will comprise the separated tracks.
Code and detailed instructions on how to reproduce experiments in part four (source separation using deep learning) can be found on the associated github repository. The separation results and the computed metrics with BSS Eval can be found on the zenodo page. Similarly, for the score-informed version, check the zenodo page.