Skip to main content
1

Get an API key

Create a key in the Geko console. Keys look like sk-tokay-…. Keep it server-side — treat it like a password. New accounts get $1.00 of free credits, and one key covers both directions.
2

Make a sound

Pick whichever feels natural — all four do the same thing.
hello.wav is 24 kHz, 16-bit PCM mono. Play it:
3

Read it back

Now the other direction — send that same file back and get the text.
That’s the whole API — text in, audio out, audio in, text out. Same key, same balance.

Install the SDK

The SDK has no runtime dependencies and ships both ESM and CommonJS with types. Requires Node 20+ (or Deno / Bun / modern browsers).

What next

Text-to-speech

Voices, streaming, the nfe dial, and output formats.

Speech-to-text

Code-switching, accepted audio, and accuracy numbers.

Use it in your app

Server-side Next.js / Express integration, done right.

Errors & retries

Handle failures cleanly in production.
First call slow? Both GPU backends scale to zero when idle, so a cold start can take tens of seconds; subsequent calls are fast. The SDK’s default 120 s timeout accounts for this. The two services warm independently.