Skip to main content

​
Write to us

contact@namuai.org

Questions, bugs, a mistake in a Hausa or Zarma example, a higher limit for a pilot. We answer in English or French.
To help us answer quickly, include:
  • the request_id from the error (or the X-Request-Id header), and the time with your time zone;
  • the endpoint and the options you sent (not your key, and not audio of anyone who did not agree to share it);
  • what you expected, and what happened.

​
Is the API up?

/v1/health needs no key. If it answers but your requests fail, the error’s type usually says why: see errors.

​
Common questions

No. When the API has been quiet, the first request wakes the model, which can take a minute or more. The next requests are much faster. Keep generous timeouts.
Not with your key in the app. Call it from your server, and let the app call your server. See API keys.
No. The API speaks with the voices listed in voices, each used with its speaker’s consent.
Not in the API yet. They are part of the Lab’s research; we add a language when we can measure that it works.
During the beta, write to us with what you are building and the volume you expect.
Namu keeps usage records (which model, when, how much audio, how long it took, the status), not your audio or text. Speech to text and text to speech run on Namu’s own models. Translation and the voice agent’s replies are written by Google Gemini through OpenRouter, which receives the text they work on (and, for spoken French in a translation, the recording). See the privacy policy.
From the console’s Settings, write to us from your account’s email address: we delete the account, its keys and its usage records.