Masato Hagiwara

Recto — a truly 2D language Recto is a 2D programming language that uses nested rectangles as its core syntax, encoding structure and recursion directly in space instead of a linear stream of text. Recto explores new ways to wri...

Recto — a truly 2D language

masatohagiwara.net/recto.html

What if your code lived in two dimensions instead of a flat stream of text? Recto is a 2D programming language that uses nested rectangles to encode structure and recursion directly in space.

📢 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 👇🏽

@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.

NatureLM-audio: an Audio-Language Foundation Model for Bioacoustics Large language models (LLMs) prompted with text and audio have achieved state-of-the-art performance across various auditory tasks, including speech, music, and general audio, showing emergent...

Yes, and we also have an updated camera ready version here on OpenReview openreview.net/forum?id=hJV...

AVES has already been used in many bioacoustics studies— identifying sounds in the North Sea www.frontiersin.org/journals/rem..., beluga whales pubs.aip.org/asa/jasa/art..., bats arxiv.org/abs/2409.12634 to name a few. We are deeply grateful for resources like xeno-canto that made AVES possible!

I'll be at #ICLR2025 in Singapore this Saturday for poster session 6, co-presenting our NatureLM-audio project! If you're around and interested in AI for bioacoustics, ecology, or related areas, would love to meet up — feel free to reach out!

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...

AVES is now pip-installable 🎉

This self-supervised, transformer-based model is pretrained on large-scale animal vocalization datasets & thanks to our incredible engineering team is now more accessible than ever–ready to run with just a single command.

🔗 github.com/earthspecies...

We’ve grown our research team!

This high-caliber group, led by Chief Scientist Olivier Pietquin, is already working hard to scale our models & projects and drive our progress toward a future of interspecies understanding 🔬🌍

Learn more about our work: bit.ly/4iMmNLE

Two Years Later: The Most Important Cancer Lesson When I was first diagnosed with lung cancer, I thought I had only a few months to live. But two years later, I’ve learned one of the most important lessons: don’t fixate on survival statistics. Here's...

Today marks two years since my lung cancer diagnosis. I'm incredibly grateful to have made it this far and thankful for the support of my family, friends, and colleagues. I wrote about the most important lesson I’ve learned on this journey:

www.resnres.org/20250317-mos...

Biodenoising—A Novel Method for Noise Reduction In Animal Vocalizations - Earth Species Project Biodenoising, a novel method for animal vocalization denoising that does not require access to clean data, was recently accepted to ICASSP 2025. Read more about it and start using it today.

One of our latest papers, Biodenoising, has been accepted to #ICASSP2025!

Biodenoising is our newest method to tackle one of ethology & bioacoustics' biggest data analysis challenges – background noise! 🧵👇🏻

www.earthspecies.org/blog/biodeno...

🎉 Exciting news! Our paper on NatureLM-audio, the first large audio-language model for understanding animal sounds, has been accepted at ICLR 2025! 🐦🔊

A big step forward in using AI to support biodiversity and better understand the natural world.

Check out the paper below!

Introduction to Japanese Natural Language Processing A thorough guide to engineering problems that come with working with Japanese, from basics like tokenization through cutting edge techniques.

Our book, Introduction to Japanese Natural Language Processing (co-authored with Paul x.com/polm23), now includes a new section "6.3 Sequence Labeling with BERT"! Learn how to build an NER system with BERT.

The book is nearing completion—stay tuned! 📚✨ www.japanesenlp.com

Is Sugar Bad for Cancer? I covered a recently published Nature paper on dietary fructose and cancer growth. It shows fructose indirectly helps tumors grow through liver lipids, while another study shows it directly feeds lung...

A recently published Nature paper shows fructose indirectly helps cancer tumors grow through liver lipids, while another study shows it directly feeds lung cancer cells. Cutting back on high-fructose foods like soda and fruits juice may help patients. www.resnres.org/20241207-is-...

Project AER - Endless [lofi hip hop/sleepy beats] YouTube video by Lofi Girl - Chill Beats

An example of Waymo-style ambient lo-fi music www.youtube.com/watch?v=skY_...

Highlight of our recent SF trip—self-driving Waymo! I was impressed by something often overlooked: their choice of ambient music and voice. So different from Uber, where you can't choose music and may have to endure driver's loud phone calls in a language you don't understand.

第67回: 手話の自然言語処理 YouTube video by NLPコロキウム

Just watched @kayoyin.bsky.social's fantastic talk at the Japanese NLP colloquium on sign language processing. It broadened my perspective on what language is and how AI can support Deaf and hard-of-hearing communities. Highly recommended! youtu.be/7dD8wu-Chbo

Travel Smarter: The Key to Boosting Happiness on Vacation We’re heading into the holiday season, and I’ve come across research highlighting the wellness benefits of travel, including for those with health challenges. Studies show that anticipation boosts hap...

As we head into the holiday season, I found research on travel's wellness benefits, even for those with health challenges. Studies show that anticipation boosts happiness, relaxed trips are more effective, and shorter, frequent trips are better. www.resnres.org/20241122-tra...

🎙️✨ Join me and the researchers behind NatureLM-audio (Marius Miron and David Robinson) as we present a live technical walkthrough on Nov 21 at 5pm GMT (12pm US Eastern / 9am US Pacific)!

Interested? Head to our Discord community for more information: discord.com/invite/H2Y53...

Axolotl: Fine-Tune and Instruct-Tune LLMs with Ease Axolotl is an easy-to-use framework for fine-tuning and instruction-tuning LLMs. It supports modern techniques like LoRA and quantization, and simplifies instruction-tuning with clear configuration op...

I've been using Axolotl to quickly run experiments with fine-tuning and instruction-tuning for LLMs. Highly recommend for anyone looking to simplify their process! www.resnres.org/20241115-axo...