Managing the Music Technology Promotion Office of Sony Computer Science Laboratories Previous : Head of innovation @ Sony Music Entertainment France
Joined April 2009
- Tweets1.1K
- Following983
- Followers753
- Likes973
My six-year-old told me: "ask ChatGPT." I thought I had to do something.
10 months later: a robot in the living room, a brain in a box, and a 60-page honest account. What worked, what broke, what I got wrong.
No episodes. Read it or don't.
medium.com/@michaelturbot/my…
michael turbot retweeted
I sent ChatGPT an audio file of a series of FART sound effects and asked what it thinks of "my music" and this is what it said
michael turbot retweeted
📽️ Join us for our #seminar with #MarcoPasini, PhD student at #QueenMaryUniversity of London
In collaboration with @SonyCSLMusic he researches ways to make generative models for audio and music both faster and more controllable
⚡Don’t miss it: youtube.com/live/g-9JOEFM6ck
michael turbot retweeted
😃Accepted @ieeeICASSP papers of @SonyCSLMusic:
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 Embedding Learning
H. Tian, S. Lattner, B. McFee, C. Saitis
Music2Latent2: Audio Compression with Summary Embeddings and Autoregressive Decoding
M. Pasini, S. Lattner, G. Fazekas
Zero-shot Musical Stem Retrieval with Joint-Embedding Predictive Architectures
A. Riou, S. Lattner, A. Gagneré, G. Hadjeres, S. Lattner, G. Peeters
Congrats to the authors!
@latentspaces @howariou @GiorgiaCanti @tiianhk @marco_ppasini @GeoffroyPeeters @gaetan_hadjeres @SonyCSLParis
Hey @Garmin , for 2025 could we be able to see when - on the load page - be able to see what activity changed the « load »just by clicking on the dot or the line ?
michael turbot retweeted
💡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
michael turbot retweeted
🥳 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 😉).
@SonyCSLMusic
michael turbot retweeted
📡We'll be in #Cannes all day on November 29 to talk about market gardening. Come and join us for #Symposium
🪴What models for small-scale market gardening?
🌐The conference will also be broadcast online on our networks ➡️ free registration required: docs.google.com/forms/d/e/1F…
Super moment à Kigali ! Bravo !
🌍 Lors de #Acces2024, j’ai eu l’opportunité de participer à un panel dédié aux innovations et opportunités dans l’industrie musicale africaine et mondiale aux côtés de Thomas Zandrowicz, Esther NAAH, @mturbot et Kobby Ankomah Graham.
michael turbot retweeted
A list of diffusion works & tutorial (at #ISMIR2024) from our lab!
[ML]
arxiv.org/abs/2405.17251 #NeurIPS24 (GenWarp: Novel View Synthesis)
arxiv.org/abs/2405.14822 #NeurIPS24 (PaGoDA: Multi-Scale 1 Step Generator)
arxiv.org/abs/2310.02279 #ICLR24 (CTM: Fast Image Gen.)
arxiv.org/abs/2311.16424 #ICLR24 (MPGD: Guided Diffusion)
arxiv.org/abs/2210.04296 #ICML23 (FP-Diff: Consistency-type Model)
arxiv.org/abs/2301.12686 #ICML23 (Blind Inverse)
[Audio/NLP]
arxiv.org/abs/2402.17011 #ACL24 (Knowledge Gen.)
arxiv.org/abs/2402.06178 #IJCAI24 (Music Editing)
arxiv.org/abs/2309.06934 #ICASSP24 (Declipping)
arxiv.org/abs/2305.10734 #ICASSP24 (Speech Enh.)
arxiv.org/abs/2210.05148 #ICASSP23 (Music Transcription)
arxiv.org/abs/2210.07508 #ICASSP23 (Vocoder)
arxiv.org/abs/2211.04124 #ICASSP23 (Dereverb)
arxiv.org/abs/2210.17287 #INTERSPEECH23 (Speech Enh.)
michael turbot retweeted
``Music Foundation Model as Generic Booster for Music Downstream Tasks,'' WeiHsiang Liao, Yuhta Takida, Yukara Ikemiya, Zhi Zhong, Chieh-Hsin Lai, Giorgio Fabbro, Kazuki Shimada, Keisuke Toyama, Kinwai Cheuk, Marco Martinez, Shusuke Takahashi, Stefan Uhl… ift.tt/1POz4NE
michael turbot retweeted
Music Foundation Model as Generic Booster for Music Downstream Tasks. arxiv.org/abs/2411.01135
michael turbot retweeted
Our #MusicTeam just released its 1st AI-prototype: #SampleMatch !
Try it now on @TechHubbySony: samplematch.csl.sony.fr/
Congrats Team 👏
#techhubsony #sonycslparis #aimusic #airesearch #musicprototype
🎶Discover #SampleMatch, our first public AI prototype designed to facilitate your music production experience
Try it today @TechHubbySony: samplematch.csl.sony.fr/
🔍Find perfect drum samples instantly
🎨Explore new creative ideas
⚡Seamlessly integrate into your workflow
michael turbot retweeted
🌱Midori Farm: where tradition meets technology in urban agriculture
🌿We’re combining the time-honored #FrenchMethod market gardening with modern tools to bring sustainable farming to urban spaces
Our goal is to explore the future of small-scale microfarms in peri-urban areas
michael turbot retweeted
😯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 @ISMIRConf paper, where we use Information Content curves as a control signal for polyphonic music generation:
📜arxiv.org/pdf/2408.06022
Great work by @BjareMathias !🤓 @SonyCSLMusic @cpjku @SonyCSLParis
michael turbot retweeted
🎶New paper alert!
📜#MaartenGrachten & @latentspaces introduce Accompaniment Prompt Adherence (APA)—a new metric for evaluating AI-generated musical accompaniments.
➡️arxiv.org/abs/2404.00775
⚙️github.com/SonyCSLParis/audi……
#SonyCSLParis #AI #MusicTech #CLAP
michael turbot retweeted
🎼 To keep your Summer entertained, we're releasing a new Reggae demo produced by #CarliNistal using our AI model #DiffARiff
We're working on more demos in multiple musical styles for you to listen to before heading back to work!
#aimusic #ai #research #innovation #sonycsl
michael turbot retweeted
🥳Another ISMIR Paper released (+model weights)!!
Music2Latent: Consistency Autoencoders for Latent Audio Compression
Marco Pasini, Stefan Lattner, George Fazekas
64x compression, 48kHz, HQ reconstruction
- End-to-end training with only one loss term
- Representations competitive in downstream tasks
- First Consistency Autoencoder (in any domain)
📜 Paper: arxiv.org/abs/2408.06500
🌍 Code + Weights: github.com/SonyCSLParis/musi…
@ISMIRConf @SonyCSLMusic
More below 👇
🔊 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/musi…
Paper: arxiv.org/abs/2408.06500
How? 👇🧵
