Text-to-speech
Text to speech that sounds human.
Voices with real range: emotion, emphasis, and pacing that land the way you mean them. Type any line and hear it read back in a voice you would believe.
Miles
British
Cross-lingual transfer
Every voice. Every language.
Every preset voice speaks all 10 output languages, carrying its own identity across each one. Fidelity is strongest in the voice's native language.
Englishen
Speak the name, the old man whispered, and the door will open. But say it softly. Some things beneath this hill have been listening for a very long time.
For developers
One OpenAI-compatible endpoint.
The same voices you just heard are a single POST away. If you already call OpenAI, you are three lines from calling Sprag.
from openai import OpenAI
client = OpenAI(
api_key="sk-sprag-...",
base_url="https://api.sprag.ai/v1",
)
response = client.audio.speech.create(
model="chorus-voices",
voice="sienna",
input="The train leaves at nine. Please do not be late.",
response_format="mp3",
)
with open("hello.mp3", "wb") as f:
f.write(response.content)- Create an API keyGenerate one in the dashboard. It scopes to your account and its usage.
- Point the base URL at SpragSet base_url to https://api.sprag.ai/v1. Nothing else in your client changes.
- Call /v1/audio/speechSend text and a voice ID. Get mp3, wav, or opus back in one round trip.
Drop-in OpenAI
The official client works unmodified.
10+ languages
Every voice, cross-lingual.
mp3, wav, opus, + more
Pick the output format per request.
56+ preset voices
Plus voices you clone or design.