The preprint of our #ICML2021 paper, Relative Positional Encoding for Transformers with Linear Complexity, is out! Thanks to Antoine Liutkus, Shih-Lun (Sean) Wu, Umut U. Simsekli, Yi-Hsuan Yang and Gaël Richard. MIP-Frontiers Marie Skłodowska-Curie Actions arxiv.org/abs/2105.08399

We've been working on extending our NeuroDrum to generate drum loops 🥁! It's called LoopNet and it was accepted for IEEE ICASSP. The paper is now on IEEEXplore and we created an interactive demo on Colab. 📝Paper: ieeexplore.ieee.org/document/94150… Demo and Code: github.com/aframires/drum…

Join us in the live discussion of our latest lyrics alignment paper at #ICASSP2021 on Thursday, June 10 at 7pm BST / 2pm EDT. "Low Resource Audio-to-Lyrics Alignment from Polyphonic Music Recordings" ieeexplore.ieee.org/document/94143… #ICASSP2021 MIP-Frontiers C4DM at QMUL

Today I'll be presenting our work on 🥁drum loop🥁generation on IEEE ICASSP! If you are there, stop by the Applications in Music and Audio Processing Session at 22h30 Spanish time. Presentation+Poster: ieeeicassp-virtual.org/presentation/p…

I'm presenting our work on "identifying setlists from long audio queries" today at 8pm CEST at #ICASSP21. Come and find out whether automating this task is realistic or not. Paper: bit.ly/setlist21-paper Video: bit.ly/setlist21-video w/ Emilio Molina Joan Serrà Emilia Gomez

Embracing the virtual IEEE ICASSP 2021 on gather.town! Horizon 2020 Marie Skłodowska-Curie Actions EU Science & Innovation 🇪🇺 #MSCA #ResearchImpactEU #H2020

📝We transcribed the lyrics of MUSDB18, a widely used dataset for music source separation. You can use them for your research on automatic lyrics transcription, informed source separation, etc. An IEEE TASLP paper using the data will be published soon. zenodo.org/record/3989267

I feel delighted to announce that our latest work “MSTRE-Net: Multistreaming Acoustic Modeling for Automatic Lyrics Transcription” is accepted to be published at ISMIR2021. We introduce a novel NN architecture and report the new state of the art. @ismir2021 MIP-Frontiers C4DM at QMUL

We've been working on a video that briefly explains the tasks of MIR! You can now see it on the following link: youtube.com/watch?v=YgYV-7…

Some weeks ago, our paper "DarkGAN: Exploiting Knowledge Distillation for Comprehensible Audio Synthesis With GANs" was accepted @ismir2021🤠 📑arxiv.org/abs/2108.01216 🔊an-1673.github.io/DarkGAN.io/ 💻github.com/SonyCSLParis/D… Stefan Lattner SonyCSL(Paris)_Music Team Télécom Paris MIP-Frontiers

🗣Informed Singing Voice Separation. 📝Lyrics Alignment. ↗️Monotonic Attention Mechanism. I am happy to share our IEEE/ACM TASLP paper where we make a small contribution to these topics: ieeexplore.ieee.org/abstract/docum… Open access version: hal.telecom-paris.fr/hal-03255334/d…

Antoine Liutkus's #ICML2021 presentation about our Stochastic Positional Encoding (SPE) is now online! Paper: arxiv.org/abs/2105.08399 Yi-Hsuan Yang Shih-Lun (Sean) Wu Umut U. Simsekli MIP-Frontiers ICML Conference slideslive.com/38958574

Our PhD student Ondřej Ondřej Cífka has been awarded the 🏆 Prix Jeune Chercheuse/Chercheur of Science & Musique, which will take place this Saturday, Oct 9 at @DiapasonRennes! Ondřej will present his work on #music #styletransfer. adasp.telecom-paris.fr/news/highlight…

MIP-Frontiers project has supported 15 PhD students. We will have a final workshop for the fellows to present the main results of their research. We have also invited several distinguished researchers in the field to give keynote presentations. Programme: mip-frontiers.eu/2021/10/11/fin…

Our final workshop has started! Today we will have 9 presentations by the fellows, as well as a keynote by Augusto Sarti: Applications of Machine Intelligence in Computational Acoustics. Full programme and link: mip-frontiers.eu/2021/10/11/fin…

We are back to our final workshop after a lunch break. Right now we have Augusto Sarti with a keynote on "Applications of Machine Intelligence in Computational Acoustics". Programme and link: mip-frontiers.eu/2021/10/11/fin…

Last day of our final workshop! We will start with a keynote by Rachel Bittner entitled "Source Separation Metrics: What are they really capturing?". After, we will be back to the ESR presentations! Programme and link: mip-frontiers.eu/2021/10/11/fin…

Back to the MIP Frontiers Final Workshop with a keynote by Kazuyoshi Yoshii about "Generative Modeling Meets Deep Learning: A Modern Statistical Approach to Music Signal Analysis". After we will have the last presentations of the workshop. Program+link: mip-frontiers.eu/2021/10/11/fin…

Unsupervised audio source separation with deep learning and some inspiration from DDSP: arxiv.org/abs/2201.09592

We need more women in STEM to address global challenges! 'Women in STEM: Mind the Gap' Webinar on March 18th. Registration link bit.ly/3BLPOmn