First Principles podcast

EP86 - Neural Codec TTS and the Secret to Cloning a Voice

0:00
21:03
Reculer de 15 secondes
Avancer de 15 secondes
You pick up the phone and hear a loved one's voice, perfectly capturing their unique cadence and the familiar echo of their living room, only to realize it is a machine generating speech in real time. We assume this software is just a highly advanced sound editor, splicing and repitching tiny slivers of a recorded voice to trick our ears. But the truth involves a radical shift in how computers understand the world, transforming the continuous physics of sound into a foreign language that a machine can learn to speak. If you enjoy the show and want to support it, you can buy me a coffee: https://paypal.me/podcast8

D'autres épisodes de "First Principles"