More-natural prosody for synthesized speech
Amazon just solved the biggest problem in AI speech synthesis: making voices sound actually human.

Why it matters
This breakthrough in prosody transfer could accelerate enterprise adoption of AI voice assistants and automated customer service, addressing the uncanny valley problem that has limited deployment at scale.
The key facts
3 to knowProsody transfer technique reduces source speaker leakage
Prosody selection model improves semantic content matching
Published by Amazon Science research team
Go to the source
Amazon Scienceamazon.science
Publisher excerpt: Prosody transfer technique addresses the problem of “source speaker leakage”, while prosody selection model better matches prosody to semantic content.