Amazon Polly
last releaseMay 2026
powered byAmazon Polly generative engine
goblin vibe check:
not flashy, but the realtime streaming path is exactly what voice agents need in production
aws' production text-to-speech service for generating speech across many voices, languages, and deployment patterns. the new bidirectional streaming api makes it more relevant for realtime agents and game-adjacent voice prototypes because speech can be generated incrementally instead of waiting for full request-response turns.
text-to-speech service with neural and generative voice optionsbidirectional streaming api supports lower-latency interactive speech workflowsaws deployment path fits production apps and enterprise voice systemsuseful for narration, accessibility, npc prototypes, and voice-agent output
key features
text-to-speech service with neural and generative voice optionsbidirectional streaming api supports lower-latency interactive speech workflowsaws deployment path fits production apps and enterprise voice systemsuseful for narration, accessibility, npc prototypes, and voice-agent output
spec & usage
bidirectional streaming api announced May 2026
runs through AWS cloud APIs and AWS account billing
best when voice generation needs operational reliability more than boutique voice cloning
limitations
less creator-native than ElevenLabs or other specialized voice studios
aws setup and billing overhead are real for small solo projects
scope:
audiotoolvoiceapicloudpaidreal-timebusiness
launch2016
last releaseMay 2026