Amazon Polly
AWS's cloud TTS service, supporting neural and standard TTS with unique SSML tags and pronunciation lexicon features.
Amazon Polly is AWS’s cloud text-to-speech service. As one of the earliest cloud TTS products on the market, Polly has accumulated mature and stable technology. The service supports both standard TTS and neural TTS synthesis methods, covering over 30 languages and offering more than 60 voices to choose from. A standout feature of Polly is its powerful pronunciation lexicon and SSML tag support, which allows developers to customize pronunciation rules for specific words. This is particularly useful in scenarios requiring precise control over the pronunciation of proper nouns, brand names, or industry terminology.
In terms of pricing, Amazon Polly is consistent with other mainstream cloud providers, with standard and neural tiers priced at $4 and $16 per million characters respectively. However, for new users, Polly’s free tier is quite generous—5 million characters per month for the first year, far exceeding comparable offers from Azure and Google, making it ideal for early-stage development and prototyping. Polly also supports streaming and SSE, allowing audio to begin playback while still being generated, effectively reducing perceived latency.
Typical use cases for Amazon Polly include: powering content narration for smart voice assistants like Alexa, providing multilingual voiceovers for online education platforms, enabling article-to-speech functionality in news and media applications, and delivering voice interaction capabilities for IoT devices and smart hardware. As part of the AWS ecosystem, Polly integrates seamlessly with services such as Lambda, S3, and CloudFront, making it easy to build complete speech processing pipelines. However, it is worth noting that Polly offers fewer voices compared to Azure and Google, with relatively limited coverage, particularly for Asian languages.