Skip to navigation

SpeechifyAI Build

Build speech into your product with one API. Generate audio, stream long-form text, clone voices, and control delivery with SSML.

Your first request

POST
/v1/audio/speech
from speechify import Speechify
client = Speechify(
    token="YOUR_TOKEN_HERE",
)
client.audio.speech(
    audio_format="mp3",
    input="Hello! This is the Speechify text-to-speech API.",
    model="simba-3.2",
    voice_id="geffen_32",
)
Try it

Ready to run it end to end? The Quickstart walks you through your first call: get a key, install the SDK, generate speech, and play it.

Grab an API key at platform.speechify.ai/api-keys and set SPEECHIFY_API_KEY so the SDKs authenticate automatically.

Set up

Install an SDK

pip install speechify-api for Python, npm install @speechify/api for TypeScript. Both read SPEECHIFY_API_KEY from the environment automatically.

Authenticate

A single Authorization: Bearer key works for every endpoint. Manage and rotate keys in the console.

Build With Speech

Streaming

Start playback before the full audio is generated. Up to 20,000 characters per request.

Voice cloning

Clone a voice from a 10-30 second sample, with verified consent from the speaker. Cloned voices work across every supported language.

SSML and emotion

Fine-grained control over pitch, rate, pauses, emphasis, and 13 emotion presets.

Speech marks

Word-level timestamps for highlighting, captions, and audio-text sync.

Integrations

Speechify voices drop into every major voice-agent platform. Native plugins where they exist, an open-source tts-shims proxy where they don’t.

LiveKit

Add speechify.TTS(...) to a LiveKit AgentSession via the official Python plugin.

Pipecat

Use Pipecat’s built-in SpeechifyHttpTTSService as the TTS stage, with word timestamps.

Vapi

Serve Speechify PCM to Vapi custom voice via the tts-shims Vapi provider.

Deepgram

Point Deepgram Voice Agent’s open_ai speak provider at the tts-shims shim.

See the Integrations overview for the platform picker, including Puter.

Models and languages

ModelBest forLanguagesHighlights
simba-3.2Recommended for new English integrationsEnglishLowest TTFB, richest expressivity; the recommended Simba 3 model
simba-3.0Streaming-native beyond English6 languages, 7 localesThe API default when model is omitted; English, German, Spanish, French, Italian and Brazilian Portuguese, set language to pick one
simba-english, simba-multilingualRetired (Simba 1.6)A new workspace gets 400 model_retired: use simba-3.2 for English and simba-3.0 for German, Spanish, French, Italian or Portuguese. For any other language, check each model’s languages in GET /v1/audio/models. Only a workspace pinned to an API version before 2026-09-21 can still name them, and from 2026-11-21 they are served by our current models

See Models and Language Support for the full matrix. Still choosing a vendor? Best TTS APIs of 2026 puts Simba 3.2 beside the other models it covers, all evaluated on the Artificial Analysis Speech Arena, with each one’s price per million characters.

Resources

API Reference

Endpoint schemas, parameters, and response shapes.

Examples

End-to-end demo projects on GitHub.

Console

Manage API keys, voices, and billing.