streaming text to speech

Start playback in under 300ms for responsive, real-time voice.

Sub-300ms first-byte latency
Chunked audio streaming
u-law and PCM output
Built for real-time apps

definition

Streaming text-to-speech returns audio in chunks as it is synthesized, rather than making the caller wait for the whole clip. The first byte arrives in under 300ms, so playback begins almost immediately. It is the mode real-time experiences depend on.

why-latency-matters

In a live experience, a delay before speech is a dead moment. A voice agent that pauses feels broken; a read-aloud button that hangs feels slow. Sub-300ms first-byte latency keeps the interaction feeling immediate and natural.

how-it-works

The API streams chunked audio over the connection as it generates, so the client plays the opening while the rest is still being produced. This overlap is what removes the wait that a full render imposes.

formats

Streaming output includes u-law for telephony and PCM for media pipelines, so the same low-latency stream feeds phone systems, voice agents, and apps directly. See the streaming guide for details.

FAQ

Frequently asked questions

What is streaming text-to-speech?

Streaming text-to-speech returns audio in chunks as it is generated instead of waiting for the full render. The first audio arrives in under 300ms, so playback starts almost immediately, which is what real-time apps like voice agents and read-aloud features need.

How low is the latency?

First-byte latency is under 300ms, so a listener hears speech almost as soon as the request is sent. That responsiveness is the difference between a natural conversation and an awkward pause.

What output formats does streaming support?

Streaming supports formats including u-law for telephony and PCM, so the audio drops into phone systems, voice agents, and media pipelines without a conversion step.

speech marks multilingual text to speech streaming text-to-speech realtime text-to-speech what is text to speech

Start building

Start playback in under 300ms for responsive, real-time voice.

Get an API key, free Talk to sales