Breaking02 / IT
AWS rolls out Amazon Polly voice cloning with limited access
Customers must confirm the speaker’s consent before creating a cloned voice, and AWS charges for speech synthesis based on usage.
Amazon Web Services (AWS) announced a limited release of voice cloning for its text-to-speech service Amazon Polly on 20 Nov 2026, with the feature available in AWS regions that support Polly’s generative engine.
AWS said some customers already have access and others can apply. Creating a cloned voice is free, and AWS charges for speech synthesis based on usage.
The list AWS published omits Korean. Each cloned voice speaks only the language used in the recording, according to AWS.
AWS said Polly creates cloned voices from recordings of consenting speakers. Customers must confirm the speaker’s consent each time they create a voice. Polly adds a watermark to speech generated with a cloned voice.
AWS said customers can select a cloned voice when using Polly’s existing speech synthesis features. Only the AWS account that created the voice can use it.
AWS did not disclose in its announcement how to detect the watermark in generated audio.