We've improved the SOTA of singer representation learning (comparing different SSL techniques). We provide paper, code, and trained models: 🪩Blog: tinyurl.com/ssl-singer 📜Paper: tinyurl.com/ssl-sing-paper 🧾Code: tinyurl.com/ssl-sing-github Great work Bernardo Torres! 😃 ISMIR Conference

New paper: PESTO pitch estimation is on par with CREPE, but ~800x smaller. Due to its self-supervised objective, it can be easily trained on custom data. 🪩Site: tinyurl.com/pesto-pitch 📜Paper: arxiv.org/pdf/2309.02265… 🧾Code: github.com/SonyCSLParis/p… SonyCSL(Paris)_Music Team Alain Riou

PESTO is a pitch estimator that uses a smart architecture design to naturally get pitch-shift invariance. This + SSL training results in an extremely small model with a great performance. Cool work!

Oldies but goldies: H. Kuhn, The Hungarian Method for the assignment problem, 1955. The most celebrated algorithm to solve efficiently (almost cubic time) the assignment problem. en.wikipedia.org/wiki/Hungarian…

Huan had her visa to travel to Italy for ISMIR Conference rejected - for the second time in a row, despite having papers to present both years, which is incredibly unfair! If anyone can help, please get in touch 🙏

Happy to be participating in this year’s #ISMIR2023!

Congratz Alain Riou , Stefan Lattner, Gaëtan Hadjeres , Peeters Geoffroy on the best paper award with PESTO at ISMIR Conference !!!!

the math in most diffusion papers

🥳 We present our #ICASSP2024 paper: A diffusion model that generates production-quality (bass) audio stems to any audio input. 😎🎸 According to our experience, that's more useful to artists than generating full mixes. 🙃 📜Paper: arxiv.org/abs/2402.01412 🎶Demo:

Great stuff!

So many amazing people in this photo!

Signals with a high degree of autocorrelation (such as pitched signals) make the training of convnets on raw audio unstable. For Gaussian initialization, the greater the input’s autocorrelation, the greater the variance of the output. Huge relief seeing these kinds of papers <3

Interested in ill-posed learning tasks, uncertainty prediction, conditional density estimation or multi-head deep neural networks ? In our new paper, accepted at #ICML24, we tackle these challenges by exploring the Winner-Takes-All (WTA) training scheme. [1/n]

🌟My keynote at the C4DM at QMUL workshop about "Models of Musical Signals: Representation, Learning & Generation" is now on YouTube, giving an overview on developments in self-supervised learning for audio since 2020, low-level representation learning, audio (stem) generation and much

PESTO 2.0 è rilasciato! 🥳🥳🥳 With Brazilian chef Bernardo Torres (and others), we revisit this traditional italian sauce, invented in Milan at ISMIR Conference 2023 🇮🇹 And you can taste it in REAL-TIME at home (~5 ms latency) ⏱️ 1/6

We’ve added fresh ingredients and cooking tricks to make the best, lightest, and fastest neural pitch estimator even better! 🌿🔥 Shoutout to all collaborators and to the amazing chef & SSL titan, Alain Riou

🔉New paper out! Recent audio codecs typically learn compression and quantization jointly, limiting the choice of quantization layers, non-differentiable by definition. What if we used powerful neural quantizers like Qinco2 and trained them offline? arxiv.org/abs/2503.19597