AI & TTS Setup

OpenRouter (AI)

OpenRouter gives you access to hundreds of AI models — Llama, Claude, GPT, and more — for fractions of a penny per request.

Create an account at openrouter.ai and generate an API key under Profile → Keys
Paste the key into config.lua:

Config.Keys = {
    ['openrouter'] = "YOUR_KEY_HERE",
}

Config.AI = {
    ['provider'] = "openrouter",
    ['model']    = "meta-llama/llama-3.1-8b-instruct",
}

Config.ServerWebsite = "https://yourwebsite.com"
Config.ServerName    = "YourServerName"

Leaving ServerWebsite and ServerName blank risks rate limiting or blocks from OpenRouter.

Google’s free tier covers roughly 1–4 million characters per month depending on the voice model. Most servers never hit a charge.

Go to console.cloud.google.com, create a new project, and enable the Cloud Text-to-Speech API
Under APIs & Services → Credentials, create an API key and copy it
Paste the key into config.lua:

Config.Keys = {
    ['google_tts'] = "YOUR_KEY_HERE",
}

Config.AI = {
    ['voice_model'] = "en-US-Standard-D",
    ['use_cache']   = true,
}

Google requires a billing account to enable APIs, but the free tier is generous enough that most servers never see a charge.

Language	Standard Voices	Neural/Studio Voices
English (`en`)	`en-US-Standard-D`, `en-US-Standard-F`	`en-US-Neural2-D`, `en-US-Studio-O`
Spanish (`es`)	`es-ES-Standard-A`, `es-US-Standard-B`	`es-ES-Neural2-A`, `es-US-Neural2-B`
French (`fr`)	`fr-FR-Standard-A`, `fr-CA-Standard-B`	`fr-FR-Neural2-A`, `fr-CA-Neural2-B`
German (`de`)	`de-DE-Standard-B`, `de-DE-Standard-D`	`de-DE-Neural2-B`, `de-DE-Neural2-D`
Italian (`it`)	`it-IT-Standard-A`, `it-IT-Standard-C`	`it-IT-Neural2-A`, `it-IT-Neural2-C`
Portuguese (`pt`)	`pt-BR-Standard-A`, `pt-PT-Standard-B`	`pt-BR-Neural2-A`, `pt-PT-Neural2-B`
Russian (`ru`)	`ru-RU-Standard-A`, `ru-RU-Standard-D`	`ru-RU-Wavenet-A`, `ru-RU-Wavenet-D`
Chinese (`zh`)	`cmn-CN-Standard-A`, `cmn-CN-Standard-C`	`cmn-CN-Wavenet-A`, `cmn-CN-Wavenet-C`
Japanese (`ja`)	`ja-JP-Standard-B`, `ja-JP-Standard-C`	`ja-JP-Neural2-B`, `ja-JP-Neural2-C`
Arabic (`ar`)	`ar-XA-Standard-A`, `ar-XA-Standard-B`	`ar-XA-Wavenet-A`, `ar-XA-Wavenet-B`

The voice model must match your locale. If Config.Locale = 'zh' but your voice model is en-US-Standard-D, the Google API will return silence.