Streaming TTS Guide: Low-Latency Speech Audio

Stream speech with low latency

Add low-latency streaming text-to-speech to my app: request the audio stream, consume chunks as they arrive, and start playback before generation finishes. Handle backpressure and the character limit for streamed requests.

Follow [https://docs.speechify.ai/build/streaming-tts-guide.md](https://docs.speechify.ai/build/streaming-tts-guide.md).

Streaming returns audio chunks as they are generated, so playback can start before the whole clip is ready. Use it for long text and any experience where a wait before speech would feel broken.

Speech vs stream

	Speech endpoint	Stream endpoint
Character limit (spoken)	2,000	20,000
Response	Base64 JSON + metadata	Raw audio chunks
Best for	Short, pre-rendered clips	Long text, low-latency playback

Set your API key

export SPEECHIFY_API_KEY="your-api-key-here"

Install the SDK

Python

TypeScript

pip install speechify-api

Stream audio

Call POST /v1/audio/stream. The response is a stream of raw audio chunks:

POST

/v1/audio/stream

from speechify import Speechify
client = Speechify(
    token="YOUR_TOKEN_HERE",
)
client.audio.stream(
    input="Streaming long-form audio with the Speechify API.",
    model="simba-3.2",
    voice_id="geffen_32",
)

Try it

When to stream

Reach for streaming when the first-byte wait matters: read-aloud features, in-app playback of long articles, or any interface where the user presses play and expects immediate sound. For a short, fixed clip you render once and cache, the speech endpoint is simpler.

Streaming pairs well with the streaming-native simba-3.2 model. Set model: "simba-3.2" and any English voice such as geffen_32. See Models.

Next steps

Text-to-Speech API

Request essentials and voice selection.

Speech Marks

Word-level timing for synced captions and highlighting.