AWS Upgrades Amazon Polly Generative Engine with 14 New Voices, Hindi and Finnish, and GovCloud Access
Amazon Web Services announced an upgrade to the Amazon Polly Generative text-to-speech engine on October 20, 2026, deploying a new foundation model that delivers more natural-sounding, accurate speech with low latency. The release adds 14 new voices across nine locales and brings Hindi and Finnish to the Generative engine for the first time.
The new voice roster includes Kenneth and Nate for US English, Russell and Curtis for Australian English, Kiara and Arjun for Hindi and Indian English, Tina for German, Carlos for US Spanish, Leo and Carolina for Brazilian Portuguese, Hyunduk for Korean, and Suvi for Finnish. The upgraded model is engineered to handle complex conversational content - numbers, email addresses and street addresses - more reliably, reducing the need for custom pronunciation tuning.
For existing deployments, the upgrade is a drop-in replacement: API method signatures and voice identifiers are unchanged, and full Speech Synthesis Markup Language (SSML) tag support is preserved. AWS positions the low-latency streaming API for contact center experiences, chatbots and LLM-based agentic voice applications built for real-time conversations, as well as news, media, entertainment, e-learning and accessibility workloads.
The launch also expands the Polly Generative engine to AWS GovCloud (US-West), giving public sector and regulated-industry customers access to the high-fidelity voices inside sovereign cloud boundaries. All new Generative voices are available across 11 regions: US East (N. Virginia), US West (Oregon), Europe (Frankfurt, Zurich, London), Asia Pacific (Sydney, Seoul, Singapore, Tokyo), Canada (Central), and GovCloud (US-West).
Pricing remains the sharpest buyer consideration. According to the Amazon Polly pricing page, Generative voices cost $30.00 per 1 million characters outside the free tier - 7.5 times the $4.00 rate for Standard voices and nearly double the $16.00 rate for Neural voices. The Generative free tier covers 100,000 characters per month for the first 12 months, a tenth of the Neural tier's monthly allowance, so high-volume multi-turn voice agents can accumulate significant cost quickly.
Operationally, AWS describes the streaming API as low latency, but end-to-end responsiveness in production telephony still depends on network transit and client-side audio buffering, warranting real-world testing before replacing legacy IVR speech engines. For contact center technology directors and voice agent teams, the trade is materially better conversational fidelity and two new languages with zero refactoring, priced at a premium tier.