Check out our latest blog post on using Transformers for Music Transcription: g.co/magenta/mt3 Authors: Josh Gardner, Ethan Manilow, Ian Simon, Curtis Hawthorne, Rigel Swavely, Jesse Engel

Want to push the frontier of machine-learned hyperrealistic speech modeling? The Tacotron team is hiring in Mountain View, CA!

Very excited to announce this paper, with my incredible co-first-author, Matt Shannon!: arxiv.org/abs/2212.03232 An attempt to revive generative (joint) modeling as a principled approach to un/semi-supervised learning in seq2seq problems.

End-to-end TTS can be scary 🎃 because of robustness issues, especially with hundreds of attention heads 👀. Check out our latest method to learn how to tame the monster 🧟‍♀️ without sacrificing expressiveness. 🌮🤖

tinyurl.com/22kjmk8p Researchers enhance AR Transformers in TTS to handle longer sequences without dropping words. Their Very Attentive Tacotron uses alignment mechanisms, improving robustness and matching T5 systems' naturalness while extending length generalization.

It's really cool when a simple, pretty idea basically solves a specific problem. Very Attentive Tacotron (VAT) shows how to make transformer decoders faithfully follow their conditioning input in approximately monotonic seq2seq tasks like speech synthesis, and it's crazily robust

Andrew D. Huberman, Ph.D. Most people never fully adjust to Daylight Saving Time—they just sleep less. Restore permanent Standard Time for better alignment with circadian rhythms and healthy sleep!

Today, Julian Salazar presents SequenceLayers, a framework-agnostic layer API & open-source library for creation of robust, streamable sequence models. Stop by the #ICML2025 Google booth at 3pm to learn about its design principles & JAX implementation. github.com/google/sequenc…

If you know the pain of sequence training<->inference, visit the #ICML25 Google booth 3-3:30 today (Tue.)! SequenceLayers is not just a layer library, but a new approach to defining and composing sequence architectures. Try it in Colab: colab.research.google.com/github/google/… See our report:

Huge outage crashes many of the world’s biggest apps. Multiple servers are down including Snapchat, Amazon & Alexa, Ring, Roblox, Fortnite, Robinhood, Coinbase, Venmo, etc.

Let's set the clocks back to Standard Time on Sunday morning and leave them there in the spring. I believe most people will be surprised to find that, as the days naturally "get longer" as we move away from the winter solstice toward the "longest day of the year", the summer

This is the key visual. Locking the clock is better than switching twice a year. Whether you prefer standard or daylight savings is perhaps a matter of personal preference but the data on mental health in adults and kids say standard is better.

My team in GDM Frontier AI is hiring (Mountain View). If you're a researcher interested full duplex modeling, multimodal LLMs (Gemini), modality gap, joint speech/text modeling, PGMs, streamable generative models, and representation learning for language modeling -- DM me!

Julian is hosting a student researcher on our team. He's a great host, so if this textless audio generation project sounds like it's up your alley, fill out the form linked in his thread.