Integrations - ThinnestAI

Ready-made code: Text To Speech → Integrations in the console has every snippet below, ready to copy.
Where to point them: Aero TTS speaks OpenAI’s speech API, so anything with an OpenAI text-to-speech option works when you set the base URL to https://tts.thinnest.ai/v1, the API key to a full-access key, and the model to aero-tts.

LiveKit Agents

import os
from livekit.plugins import openai

tts = openai.TTS(
    base_url="https://tts.thinnest.ai/v1",
    api_key=os.environ["THINNEST_API_KEY"],
    model="aero-tts",
    voice="anvi",
)

Pipecat

import os
from pipecat.services.openai.tts import OpenAITTSService

tts = OpenAITTSService(
    base_url="https://tts.thinnest.ai/v1",
    api_key=os.environ["THINNEST_API_KEY"],
    model="aero-tts",
    voice="anvi",
    sample_rate=24000,
)

OpenAI SDK

import os
from openai import OpenAI

client = OpenAI(base_url="https://tts.thinnest.ai/v1", api_key=os.environ["THINNEST_API_KEY"])
client.audio.speech.create(model="aero-tts", voice="anvi", input="नमस्ते!").write_to_file("hello.mp3")

import OpenAI from "openai";

const client = new OpenAI({ baseURL: "https://tts.thinnest.ai/v1", apiKey: process.env.THINNEST_API_KEY });
const speech = await client.audio.speech.create({ model: "aero-tts", voice: "anvi", input: "नमस्ते!" });

Phone calls

Use the native endpoint with "format": "mulaw", "sample_rate": 8000. The bytes can be written straight into the carrier’s media stream.