LMNT TTS

Blizzard and Aurora TTS models via the LMNT HTTP API.

LMNT TTS

LMNTTTS calls LMNT’s /v1/ai/speech/bytes endpoint. The Patter port defaults to format: "raw" (PCM_S16LE) at 16 kHz for direct telephony use; callers can switch to mp3, mulaw, wav, aac when needed.

Install

npm install getpatter

No vendor SDK is required — LMNTTTS uses the platform fetch, so it works on Node 18+.

Usage

import { LMNTTTS } from "getpatter";

const tts = new LMNTTTS();                                            // reads LMNT_API_KEY
const tts2 = new LMNTTTS({ model: "blizzard", voice: "leah" });

In an agent:

// npx tsx example.ts
import { Patter, Twilio, DeepgramSTT, LMNTTTS } from "getpatter";

const phone = new Patter({ carrier: new Twilio(), phoneNumber: "[phone]" });

const agent = phone.agent({
  stt: new DeepgramSTT(),                             // DEEPGRAM_API_KEY from env
  tts: new LMNTTTS({ model: "blizzard", voice: "leah" }),  // LMNT_API_KEY from env
  systemPrompt: "You are a helpful assistant.",
});

await phone.serve({ agent });

Options

Option	Default	Notes
apiKey	—	Reads from LMNT_API_KEY when omitted.
model	"blizzard"	"blizzard" or "aurora".
voice	"leah"	LMNT voice id.
language	undefined	Auto-derived from model when omitted.
format	"raw"	"aac", "mp3", "mulaw", "raw", "wav".
sampleRate	16000	8000, 16000, or 24000.
temperature	1.0	Higher = more expressive.
topP	0.8	Controls stability.
baseUrl	LMNT /v1/ai/speech/bytes	Override for proxying or self-hosted gateways.