Readings before audio
A plan returns the exact spoken text for every segment, with issues for anything unreadable — show it to your users or check it in your pipeline.
Mathematical speech in your product.
Learning platforms, homework tools and publishing pipelines can send text and LaTeX, show users the spoken reading, and render narration in one of two modes over a small JSON API.
What you can do
Prices you can trust
Quotes per mode come with the plan, and jobs must state the credits they expect, so a render never costs more than the quote.
Safe to retry
Job submission is idempotent: retry with the same key and the job is created, and its credits reserved, exactly once.
How it works
Check the reading. Then render.
- Create an API key in your dashboard.
- POST text and LaTeX segments to /v1/plans; read the spoken text, issues and quote.
- POST /v1/jobs with the plan, the mode and the quoted credits.
- Poll the job, then download WAV, MP3, captions or the script.
Prepare a plan
curl "https://evaria.ai/api/narration/v1/plans" \
-H "Authorization: Bearer $EVARIA_API_KEY" \
-H "Content-Type: application/json" \
-d '{"domain": "mathematics", "segments": [
{"kind": "latex", "text": "P(A \\mid B) = \\frac{P(A \\cap B)}{P(B)}"}]}'Keep API keys on your server. A key spends your account’s credits.
Try it on your own material.
Every account starts with free credits. No card required.