gandr-ai-sdk

Gandr TTS speech provider for the Vercel AI SDK.

• First audio byte in 146 ms over the open internet, 116 ms p50 first audio, server side warm
• Reads numbers, dates, order IDs and addresses correctly
• One engine speaks 23 languages with six voices
• Every render watermarked (imperceptible, detectable)
• $10 a month for one million tokens, or unlimited, unmetered stream plans from $150/mo (annual)

Free key starts at 50,000 tokens: gandr.ai

Install

npm install gandr-ai-sdk ai

Use

import { createGandr } from "gandr-ai-sdk";
import { generateSpeech } from "ai";

const gandr = createGandr(); // reads GANDR_API_KEY from the environment

const { audio } = await generateSpeech({
  model: gandr.speech("gandr-tts"),
  text: "Order number 4-2-7-1 ships on March 3rd.",
  voice: "gandr-leo",
});
// audio is a Uint8Array of a WAV file

The model id is the engine name. Pick voices with the standard voice option: gandr-ava, gandr-dane, gandr-jenny, gandr-leo, gandr-lewis, gandr-mia. Set language to any of the 23 ISO codes, speed between 0.6 and 1.5.

Docs: gandr.ai/docs