"aha, hmm, ok"
Google Soundstorm is scarily good
Human-like audio
Multiple speakers
Long dialogues
Voice cloning
100x faster
🔊 Listen to the sample below
They plan to release it to the public like ChatGPT
Prediction for uses when it's out:
⤍ YouTube Automation 2.0
⤍ Sell "voices" for voiceovers
⤍ Automated Interview Podcasts
Check out more of their demos on their research paper