The Pulse
Amazon Polly Adds 14 Voices and Generative Hindi and Finnish
Amazon Polly’s upgraded Generative engine adds 14 voices across nine locales, with Hindi and Finnish available for the first time. AWS says the model improves speech naturalness and accuracy without changing existing integrations or prices.
AI.info Team · Written with AI · fact-checked ·
Amazon Polly’s Generative engine gains 14 voices across nine locales, bringing its catalog to 57 voices across 11 languages and 25 locales. AWS says the upgraded model also improves naturalness and accuracy, including for numbers, currency, email addresses and street addresses. Generative speech in Hindi and Finnish becomes generally available for the first time.
14 voices, including Hindi and Finnish
The new voices include 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. AWS says the additions expand the service’s language and regional coverage while retaining the existing voice names and API calls.
The upgrade is generally available in 11 AWS Regions: US East (N. Virginia), US West (Oregon), Europe (Frankfurt, Zurich and London), Asia Pacific (Sydney, Seoul, Singapore and Tokyo), Canada (Central), and AWS GovCloud (US-West). AWS says the GovCloud availability gives public-sector and regulated-industry customers another deployment option.
Numbers and addresses are a stated focus
Amazon is pitching the new model for speech tasks where a misread detail can undermine an otherwise fluent exchange. Its examples include order confirmations containing tracking numbers, delivery addresses, dates, times and prices. AWS says the updated voices can also respond more closely to textual cues for delivery styles such as uncertainty or surprise, without requiring developers to configure those styles manually.
Those features target contact-center calls, voice agents, media production, e-learning and accessibility. AWS positions the low-latency streaming API as a way to send text to Polly while receiving synthesized audio at the same time, including for applications that stream responses from a large language model.
Existing integrations keep the same API and price
Developers already using Polly’s Generative engine do not need to change their code, according to AWS. The upgrade applies in place to existing generative voices, and new applications can use the same SynthesizeSpeech API. AWS says the model upgrade brings no price increase or new SKU; static synthesized prompts can still be cached and replayed without additional cost.
One limitation remains: speech marks are not available for generative voices. AWS says standard and neural voices continue to support them, and those engines are unaffected by the model upgrade. The generative engine continues to support Speech Synthesis Markup Language tags.
Sanoma says Suvi improves on its Finnish neural voice
Sanoma Media Finland uses Polly to turn real-time news articles into audio. Justus Reinikainen, Product Manager at Sanoma Media Finland, said the company publishes more than 10,000 articles a month using Polly’s Finnish neural voice and compared it with the new generative voice in testing:
“Based on our testing, Suvi’s new generative voice is a clear improvement over the previous neural voice – it sounds noticeably more natural and makes for a more enjoyable experience for the listener.”
Justus Reinikainen, Product Manager, Sanoma Media Finland
That comparison is a customer’s assessment, not a published independent benchmark. For applications that depend on speech marks, the practical choice remains specific: stay with Polly’s standard or neural engine, or use the expanded Generative catalog without speech-mark support.