Amazon Polly Adds Voice Cloning in Seven Languages

Amazon has added voice cloning to Amazon Polly, its text-to-speech service. Announced on 19 November, and summarised by the AWS News Feed, the feature is generally available as a gated release, letting approved AWS accounts create a custom synthetic voice from recordings of one consenting speaker.
Businesses can use the cloned voice for a consistent brand voice across phone systems, voice agents, mobile apps, in-store announcements, advertising and connected devices, as well as contact-centre prompts and longer content such as audiobooks, courses and podcasts.
How voice cloning works
To create a voice, a customer submits between one and ten recordings of a single speaker, with up to fifteen minutes of audio in total. AWS recommends two to ten minutes of natural speech and says quality matters more than length. The files can be uploaded through the console, included in the CreateVoice call or supplied from an Amazon S3 bucket in the target region.
The request also needs the recording language and a terms-acceptance flag. The API returns a voice ID immediately, then builds the voice asynchronously, typically in five to ten minutes, while checking speaker similarity. A finished voice can generate real-time audio or long-form audio written to S3. Because creation takes minutes, AWS advises treating it as an advance step rather than part of a live user request.
Cloned voices run on Polly's generative engine, which AWS says delivers naturalness and accuracy similar to its other generative voices. Seven languages are supported at launch: English, Spanish, German, French, Italian, Portuguese and Hindi. Each cloned voice speaks only the language in which it was recorded, and other accents of the supported languages can work but are less tested.
Consent and safety controls
Voice cloning is the kind of feature that needs guardrails, and AWS has built several in. The terms-acceptance step confirms that the customer has obtained all the rights and consents needed to create and use the voice, and AWS recommends keeping a record of consent, deleting the voice if consent is withdrawn, and telling listeners when they are hearing synthetic speech where appropriate.
Polly adds a watermark to synthesised audio made with a cloned voice. After creation, Polly does not retain its copy of the source recordings and does not use them to train Amazon Polly models. It keeps only the voice data needed for synthesis, stored encrypted until deletion, and deleting the voice removes that data immediately. Access is scoped to the creating AWS account, customers can tag voices and encrypt them with their own AWS Key Management Service key, and a voice stays in the account and region where it was created.
The recommended recording format is mono, 16-bit WAV at 24 kHz, captured from one speaker in a quiet room. AWS warns that fan noise, room echo and meeting-app audio processing can be learned by the clone, and that a cloned voice takes its delivery style from the recordings, so it cannot blend several styles into one voice.
What it costs and who can use it
Creating a cloned voice is free, while generating speech with it is charged on a pay-as-you-go basis under the standard generative engine pricing. Customers start with a quota of ten cloned voices per account per region, and the feature is available in the same AWS regions as Polly's generative engine, including AWS GovCloud in the US.
Access is gated. Select Polly customers already have it enabled, while others need to contact their AWS account team or apply, so businesses should not plan a project around it until access is confirmed. For organisations with changing scripts, the practical draw is generating new audio in an established voice without arranging a fresh recording session for every revision.