Amazon Polly Launches Bidirectional Streaming API — Real-Time TTS for Conversational AI Voice Agents

AWS Machine Learning Blog·medium signal

AWS announced a new Bidirectional Streaming API for Amazon Polly on March 27, enabling real-time text-to-speech synthesis where you can start sending text and receiving audio simultaneously. This is the infrastructure primitive voice agent builders need: stream text from an LLM and get audio output in real-time without waiting for full response generation. Competes directly with the real-time voice capabilities in Mistral's Voxtral and ElevenLabs.