iiinova — where your digital self keeps evolving

Reposted by @nkundiushuti.bsky.social

Earth Species Project (ESP) @earthspecies.bsky.social · 07/07/2026

We're excited to share that ESP’s Marius Miron is delivering a keynote at the ICML Workshop on Machine Learning for Audio on Animal Language Processing — an emerging frontier for AI and our understanding of life on Earth. @icmlconf.bsky.social 🔗 Learn more: mlforaudioworkshop.github.io

041

nkundiushuti.bsky.social @nkundiushuti.bsky.social · 30/04/2026

we are heading to ICASSP to present this paper on probing and the AVEX library: earthspecies.github.io/avex/ my colleague Gagan will be there to present the blazing fast bioacoustic data library they've been working on in the past year and plan to release soon.

earthspecies.github.io

Earth Species Project - AVEX Documentation

000

nkundiushuti.bsky.social @nkundiushuti.bsky.social · 23/04/2026

most of the work we do would be impossible without a massive engineering effort. here's to @miladalizadeh.bsky.social who's going to be at ICLR presenting our poster on bioacoustic encoders and the AVEX package. earthspecies.github.io/avex/ check this out if you work on bioacoustics.

earthspecies.github.io

Earth Species Project - AVEX Documentation

000

nkundiushuti.bsky.social @nkundiushuti.bsky.social · 20/04/2026

initially I did not know how to cope with the news, but then I learned a few important lessons from Masato the past months and one of them is that being open helps. I wrote a few things here: mariusmiron.com/blog/masato/ . Please consider supporting his family: www.gofundme.com/f/support-ma...

mariusmiron.com

Masato

000

Reposted by @nkundiushuti.bsky.social

Earth Species Project (ESP) @earthspecies.bsky.social · 16/02/2026

New preprint on vocal communication in zebra finches! 🐦 Earth Species Project and McGill University analyzed over 1.5 million female zebra finch calls to understand how female zebra finches modulate their vocalizations during natural exchanges.

1319

Reposted by @nkundiushuti.bsky.social

Earth Species Project (ESP) @earthspecies.bsky.social · 03/02/2026

Exciting news: our “What Matters for Bioacoustic Encoding” paper has been accepted to ICLR 2026! 🎉 We’ve also open-sourced the models and released AVEX, a new Python package for training and evaluating bioacoustics representation learning models. 📃 Documentation: earthspecies.github.io/avex/

022

nkundiushuti.bsky.social @nkundiushuti.bsky.social · 16/12/2025

Excited to be part of @dlbcnai.bsky.social tomorrow! I'll be presenting NatureLM-audio, @earthspecies.bsky.social 's flagship audio language model for analyzing animal vocalizations. Looking forward to great conversations with the deep learning community earthspecies.github.io/naturelm-aud...

earthspecies.github.io

NatureLM-audio: an Audio-Language Foundation Model for Bioacoustics

The world's first audio-language model designed for animal sound understanding and conservation research..

030

nkundiushuti.bsky.social @nkundiushuti.bsky.social · 28/10/2025

a few researchers from ESP @earthspecies.bsky.social will attend BioDCASE and DCASE this week in Barcelona to present three papers arxiv.org/abs/2503.00296, arxiv.org/abs/2503.02389, arxiv.org/abs/2509.04166. come say hi! we'd love to chat about our work!

000

Reposted by @nkundiushuti.bsky.social

Earth Species Project (ESP) @earthspecies.bsky.social · 16/09/2025

💡 How can we build #AI models that generalize across species + tasks? 🐦🐋🐒 This is a fundamental question for our work at ESP and it’s the topic of our most recent paper and blog post: www.earthspecies.org/blog/what-ma... 🧵Read below for more👇

earthspecies.org

What Matters for Bioacoustic Encoding: A Practical Training Recipe for Building Generalizable Models - Earth Species Project

To better understand what matters for building a generalizable bioacoustic encoder, we tested 19 models across 26 datasets and a new evaluation benchmark. Our main finding is that a two-stage training...

1113

nkundiushuti.bsky.social @nkundiushuti.bsky.social · 05/09/2025

Next week at #IBAC2025, I'll be presenting our latest work on NatureLM-audio in a poster session. 📍If you’ll be in Denmark too, come find me and Sara Keen to explore how large audio language models can be applied to bioacoustic data.

040

Reposted by @nkundiushuti.bsky.social

Earth Species Project (ESP) @earthspecies.bsky.social · 04/09/2025

🐋🐦🐸 We just launched an interactive demo of NatureLM-audio on @hf.co! 👉 Try the demo with your audio or ours, share your feedback, and help us shape the future of decoding animal communication: huggingface.co/blog/EarthSp...

0113

Reposted by @nkundiushuti.bsky.social

Earth Species Project (ESP) @earthspecies.bsky.social · 24/04/2025

📢 We've open-sourced NatureLM-audio, the first audio-language foundation model for #bioacoustics. Trained on large-scale animal vocalization, human speech & music datasets, the model enables zero-shot classification, detection & querying across diverse species & environments 👇🏽

22612

nkundiushuti.bsky.social @nkundiushuti.bsky.social · 23/04/2025

@earthspecies.bsky.social will be at @iclr-conf.bsky.social this week in Singapore to present NatureLM-audio, our foundation model for #bioacoustics. we release code, model, benchmarks earthspecies.github.io/naturelm-aud.... happy to meet folks and talk about AI for conservation, audio, LLMs.

042

Reposted by @nkundiushuti.bsky.social

Masato Hagiwara @mhagiwara.bsky.social · 15/04/2025

Our self-supervised audio encoder for animal sounds, AVES, is now pip-installable! You can install it via pip and get the embeddings from audio files with a single command. Kudos to our engineering team! github.com/earthspecies...

github.com

GitHub - earthspecies/aves: AVES: Animal Vocalization Encoder based on Self-Supervision

AVES: Animal Vocalization Encoder based on Self-Supervision - earthspecies/aves

031

nkundiushuti.bsky.social @nkundiushuti.bsky.social · 08/04/2025

I am at ICASSP and will be presenting biodenoising on Friday. happy to talk to people interested in bioacoustics,animal communication,cross-domain representation transfer or simply curious about our work at ESP.we just released a new version of biodenoising including self-training on your own data

010

nkundiushuti.bsky.social @nkundiushuti.bsky.social · 31/03/2025

encantado de participar: el jueves 3 una mesa redonda sobre ecologia sonora y genero

010

nkundiushuti.bsky.social @nkundiushuti.bsky.social · 20/02/2025

happy to announce that Biodenoising was accepted at ICASSP 2025. this is essentially the equivalent of speech enhancement for non-human vocalizations. it can be easily used in Python with pip install biodenoising.

010