Google DeepMind Audio Research: Two-Minute Multi-Speaker Dialogue

October 30, 2024

DeepMind previewed natural multi-speaker audio lasting two minutes, generated over 40 times faster than real time on one TPU and already used by NotebookLM.

Google DeepMind

Sources

Google DeepMind

Announcement · Oct 30, 2024

Pushing the frontiers of audio generation

previous event

Reuters reports OpenAI chip design work and TSMC manufacturing plans

next event

Recraft releases V3 for text and design-controlled images