Pedro Sarmento
banner
umpedronosapato.bsky.social
Pedro Sarmento
@umpedronosapato.bsky.social
270 followers 340 following 39 posts
AI & Music Data Scientist at @Music.AI | prev. @c4dm
Posts Media Videos Starter Packs
can't get enough of guitar-MIR 🎸
A new dataset (EGDB-PG) and a Tone-informed Transformer (TIT) model were developed for electric guitar transcription; TIT, trained on EGDB-PG, outperformed baselines across amplifier types due to dataset diversity and tone embedding; ablation studies assessed augmentation and embedding impact.
Towards Generalizability to Tone and Content Variations in the Transcription of Amplifier Rendered Electric Guitar Audio
Yu-Hua Chen, Yuan-Chiao Cheng, Yen-Tung Yeh, Jui-Te Wu, Jyh-Shing Roger Jang, Yi-Hsuan Yang
arxiv.org
Let us hear your AI-assisted bangers 🤘
🎶✨ AI Song Contest 2025 is HERE! ✨🎶
Are you ready to push the boundaries of music and AI?

🌍 Over 70 teams from 20+ countries have already redefined the future of music. Will YOU be next?

🔗 Learn more & sign up at www.aisongcontest.com
#AISongContest #AIinMusic #MusicInnovation #AISongContest2025
AI Song Contest
The AI Song Contest is an international competition exploring how humans can make music in collaboration with artificial intelligence
www.aisongcontest.com
Reposted by Pedro Sarmento
An exciting novel contribution by our student @jinhua-liang.bsky.social, supervised by @emmanouilb.bsky.social
Exciting research update! EECS PhD students have developed a novel approach that enables large language models (LLMs) to “hear” and “understand” sound - a breakthrough in multimodal generative #AI: www.qmul.ac.uk/eecs/news-an...
EECS PhD researcher pioneers AI that can
www.qmul.ac.uk
Thanks! I was ambivalent because I confess I didn't love the new stuff - I think it's great for the band that Manuel Gagneux let everyone write on the new album, very democratic, but 😬
How was it? Sad to miss it, specially since they seem to have played so much of the old stuff
oooops, good shout - let me get on that 🎸
Good luck to all the titans submitting to #ISMIR2025 🤘excited to see what this year's edition will bring 🎸
I'm running a paid study on guitar timbre transfer - it should take approximately 30min 🎸
If you're interested, please reach out via DM!
Reposted by Pedro Sarmento
I love how DiffRhythm keeps changing time signatures à la Dream Theater (ie, seemingly random). The vocals are in a quite deep uncanny valley, but the music sounds super good. And the audio prompting works really well! And all open source! Great job, titans <3 huggingface.co/spaces/ASLP-...
DiffRhythm - a Hugging Face Space by ASLP-lab
Blazingly Fast and Embarrassingly Simple Song Generation
huggingface.co
They're out 🤘
ISMIR 2024 Conference Proceedings are now online! ismir.net/conferences/...

Thank you to all of the authors, reviewers, meta-reviewers, and conference organizers for their contributions to a vibrant and innovative research community!

#ISMIR2024 #MIR #Music #Research
Dataset papers are usually easy to follow, and they are at the very beginning of every AI endeavour!
Reposted by Pedro Sarmento
Great interview with @jascha.sohldickstein.com about diffusion models! This is the first in a series: similar interviews with Yang Song and yours truly will follow soon.

(One of these is not like the others -- both of them basically invented the field, and I occasionally write a blog post 🥲)
History of Diffusion - Jascha Sohl-Dickstein
YouTube video by Bain Capital Ventures
www.youtube.com
Is it that good? 👀 I'm curious to try it out!
Very excited to share our latest work, the GigaMIDI dataset with > 1.4M files, published at #TISMIR 🤘 It was a huge pleasure to collaborate with such a team of titans

transactions.ismir.net/articles/10....
The GigaMIDI Dataset with Features for Expressive Music Performance Detection | Transactions of the International Society for Music Information Retrieval
The Transactions of the International Society for Music Information Retrieval publishes novel scientific research in the field of music information retrieval (MIR), an interdisciplinary research area concerned with processing, analysing, organising and accessing music information. We welcome submissions from a wide range of disciplines, including computer science, musicology, cognitive science, library & information science and electrical engineering.TISMIR was established to complement the widely cited ISMIR conference proceedings and provide a vehicle for the dissemination of the highest quality and most substantial scientific research in MIR. TISMIR retains the Open Access model of the ISMIR Conference proceedings, providing rapid access, free of charge, to all journal content. In order to encourage reproducibility of the published research papers, we provide facilities for archiving the software and data used in the research. To avoid excessive cost to the authors or their institutions, TISMIR is published in electronic-only format.
transactions.ismir.net
Reposted by Pedro Sarmento
From the 25th February to 4th March 2025, two C4DM researchers will participate at the 39th Annual AAAI Conference on Artificial Intelligence (AAAI 2025). More info at:
www.c4dm.eecs.qmul.ac.uk/news/2025-02...
The following works were authored/coauthored by C4DM PhD students and academic staff:
www.c4dm.eecs.qmul.ac.uk
this is pricelessly sad and great at the same time 🤘 Courtney LaPlante is such a titan