Glad to announce that Stem-JEPA has been accepted to ISMIR Conference ! In this work, we tackle the task of musical stem compatibility estimation (what “fits” together) as a representation learning problem. (1/7) Paper: arxiv.org/abs/2408.02514 Code: github.com/SonyCSLParis/S…

🔊 Encode and decode audio to/from latents with Music2Latent! 🔊 Music2Latent encodes only ~10 latents per second of audio 👀 This means lightning-fast training/inference of latent generative models ⚡️️ Try it: github.com/SonyCSLParis/m… Paper: arxiv.org/abs/2408.06500 How? 👇🧵

👩‍🎓SonyCSL(Paris)_Music Team has an open 3-year PhD position starting in spring 2025! We are excited to invite applicants who meet most of the following requirements: - Master's degree - Strong background in (generative) deep learning - Experience with deep learning projects - Publications

We release Accompaniment Prompt Adherence (APA), a new metric for evaluating generative musical accompaniment systems. 📜 arxiv.org/abs/2404.00775 💻 github.com/SonyCSLParis/a… The metric is training-free, uses FAD and CLAP, and correlates with human ratings of musical adherence!

🎶New paper alert! 📜#MaartenGrachten & Javier Nistal introduce Accompaniment Prompt Adherence (APA)—a new metric for evaluating AI-generated musical accompaniments. ➡️arxiv.org/abs/2404.00775 ⚙️github.com/SonyCSLParis/a…… #SonyCSLParis #AI #MusicTech #CLAP

😯Surprisal in music is an often underestimated concept. It's not only valuable for music analysis but can also be leveraged to control complexity and structure in music generation. Check out our ISMIR Conference paper, where we use Information Content curves as a control signal for

🎶Discover #SampleMatch, our first public AI prototype designed to facilitate your music production experience Try it today Sony | Technology Hub: samplematch.csl.sony.fr 🔍Find perfect drum samples instantly 🎨Explore new creative ideas ⚡Seamlessly integrate into your workflow

🎉New #DiffARiff audio demo 🎚️We continue to introduce you to Diff-A-Riff's capabilities, our accompaniment co-creation tool, with this new #Disco Demo produced by Carli Nistal 👏 Find the HQ video & audio on our YouTube channel #aimusic #research #innovation #sonycslparis

🚨 New Video Drop! 🚨 My brother, Carli Nistal, used Diff-A-Riff to create this banger! 🎶🤖 Watch it in action!

🤩 It's exciting to see how Diff-A-Riff makes it easy to build up a song based on some input audio! 👇

🌟 Exciting news! The YouTube channel ⚡AI Search⚡ just released a review of #DiffARiff, and it's already making waves—10k views in less than 12 hours and more than 100 comments with fantastic feedback! 🎉 Thanks so much for the incredible exposure!

🥳 New publication announcement! Marco Pasini solved the problem of error accumulation in continuous autoregressive models (CAMs), making it possible to generate sequences without the need for prior tokenization. Say goodbye to RVQ codecs (use music2latent 😉). SonyCSL(Paris)_Music Team

Happy to announce that our legendary DrumGAN, developed 2020 by Javier Nistal Javier Nistal is now available as a web application. Free to use and without registration! 🥁 Give it a try at drumgan.csl.sony.fr #DrumGAN SonyCSL(Paris)_Music Team

💡If you missed it, we released our new AI-tool, #DrumGAN, which allows you to generate flexible drum sounds with ease Here is a full blogpost for a detailed look at it + free download the 1st #AI-drum kits made by #Twenty9 during our collaboration ➡️🎁 tinyurl.com/blpodr

😃Accepted IEEE ICASSP papers of SonyCSL(Paris)_Music Team: Accompaniment Prompt Adherence: A Measure for Evaluating Music Accompaniment Systems M. Grachten, J. Nistal Estimating Musical Surprisal in Audio M. Bjare, G. Cantisani, S. Lattner and G. Widmer Hybrid Losses for Hierarchical

🌟 New IEEE ICASSP Paper Announcement 🌟 We introduce "Music2Latent2: Audio Compression with Summary Embeddings and Autoregressive Decoding", a novel autoencoder achieving higher fidelity at extreme compression rates using consistency models & summary embeddings. 🧵

🎭💻 We’re participating in #CultTech Residency — a new program supporting artists and creative teams exploring performance & technology Experience the benefits of working with our AI-powered tool #DiffARiff Feel free to share the news 🔽 Open call: culttech.at/residence-open…

Hey! Our paper, 🌸 “LiLAC: A Lightweight Latent ControlNet for Musical Audio Generation” got accepted as ISMIR 2025! It presents a novel control paradigm for audio diffusion models with greatly reduced parameters, encouraging users to train individual, modular models.

The deadline has been extended to June 22nd, 2025 📅 (earlier application is encouraged). Yes — you, the one reading this! If this opportunity sparks your interest, we would be delighted to receive your application.📨 We currently have the most availability in themes related to

❌ We don’t need no negative samples ❌ We don’t need no large batches ❌ No modality gap in the classroom Very happy to introduce SLAP, our latest brick in the wall of multimodal SSL 🎶🧠 Joint work with King @Juj_guinot, accepted at #ISMIR2025! 🇰🇷 1/7