What TTS Providers Are Available for OpenMAIC: Complete List and Setup Guide

OpenMAIC supports 10 TTS providers including OpenAI, Azure Speech, ElevenLabs, and a browser-native Web Speech API option, all routed through the generateTTS() factory function based on the providerId field in TTSModelConfig.

OpenMAIC implements a pluggable Text-to-Speech (TTS) framework that dynamically routes synthesis requests to cloud services or client-side engines. The system selects the appropriate implementation at runtime using a switch-based factory pattern defined in lib/audio/tts-providers.ts, allowing seamless switching between commercial APIs and local processing.

Complete List of OpenMAIC TTS Providers

The provider registry is enumerated in the header comments of lib/audio/tts-providers.ts (lines 7-15), with individual implementations handled in a switch statement beginning at line 44. OpenMAIC currently integrates the following TTS providers:

• openai-tts: OpenAI TTS API supporting voices like alloy and models including gpt-4o-mini-tts (implemented at lines 44-46).
• azure-tts: Microsoft Azure Speech Service for enterprise-grade synthesis (lines 47-49).
• glm-tts: Alibaba GLM TTS provider optimized for Chinese and multilingual generation (lines 50-52).
• qwen-tts: DashScope Qwen TTS service tailored for conversational AI applications (lines 53-55).
• voxcpm-tts: VoxCPM TTS with support for both internal and external backends (lines 56-58).
• minimax-tts: MiniMax TTS API integration (lines 59-61).
• doubao-tts: Volcengine Doubao (ByteDance) speech synthesis service (lines 62-64).
• elevenlabs-tts: ElevenLabs high-fidelity voice cloning and generation platform (lines 64-66).
• lemonade-tts: OpenAI-compatible endpoint provider for custom TTS deployments (lines 66-68).
• browser-native-tts: Client-side Web Speech API implementation requiring no external credentials (lines 69-71).

How to Configure and Use TTS Providers

To generate speech, import the generateTTS function from lib/audio/tts-providers.ts. This factory accepts a TTSModelConfig object and returns a Promise<TTSGenerationResult> containing raw audio bytes.

Explicit Provider Configuration

Define the providerId and credentials directly in the configuration object:

import { generateTTS } from '@/lib/audio/tts-providers';

const openaiConfig = {
  providerId: 'openai-tts',
  apiKey: 'YOUR_OPENAI_API_KEY',
  voice: 'alloy',
  modelId: 'gpt-4o-mini-tts',
  speed: 1.0,
};

const result = await generateTTS(openaiConfig, 'Hello, OpenMAIC!');
console.log('Audio bytes length:', result.audio.length, 'Format:', result.format);

Using Stored UI Configuration

For applications utilizing the built-in settings store, retrieve the active configuration with getCurrentTTSConfig():

import { generateTTS, getCurrentTTSConfig } from '@/lib/audio/tts-providers';

const cfg = await getCurrentTTSConfig(); // Pulls provider/voice/speed from settings
const result = await generateTTS(cfg, '欢迎使用 OpenMAIC');

// Result conforms to TTSGenerationResult interface
// { audio: Uint8Array, format: string }

Key Implementation Files

The TTS ecosystem is organized across these source files in the THU-MAIC/OpenMAIC repository:

• lib/audio/tts-providers.ts: Central factory defining generateTTS() and the provider switch logic (lines 44-71). Imports metadata from constants.ts.
• lib/audio/constants.ts: Exports TTS_PROVIDERS constant containing default URLs, required API-key flags, and available voice lists for each provider.
• lib/audio/types.ts: TypeScript definitions including TTSModelConfig, TTSGenerationResult, and provider type guards.
• components/settings/tts-settings.tsx: React component for provider selection, API key entry, and voice/speed configuration.
• hooks/use-discussion-tts.ts: Runtime hook integrating TTS generation into agent discussions.

Summary

• OpenMAIC provides 10 built-in TTS providers ranging from cloud APIs (OpenAI, Azure, ElevenLabs) to client-side synthesis (browser-native Web Speech API).
• Provider selection is controlled by the providerId field in TTSModelConfig, processed by the generateTTS() factory in lib/audio/tts-providers.ts.
• The framework returns standardized TTSGenerationResult objects containing Uint8Array audio data and format strings like 'mp3' or 'wav'.
• Configuration can be passed explicitly or retrieved from the UI settings store via getCurrentTTSConfig().
• Only the browser-native-tts provider operates without external API credentials.

Frequently Asked Questions

How do I switch between TTS providers in OpenMAIC?

Change the providerId value in your TTSModelConfig object before calling generateTTS(). Valid values include 'openai-tts', 'elevenlabs-tts', and 'browser-native-tts'. The factory function automatically routes to the correct implementation based on this identifier as defined in the switch statement at lib/audio/tts-providers.ts.

Which OpenMAIC TTS provider does not require an API key?

The browser-native-tts provider uses the browser's built-in Web Speech API and requires no external authentication. All other providers—including OpenAI, Azure, and ElevenLabs—require valid API keys configured in the apiKey field of the configuration object.

What audio format does OpenMAIC return from TTS generation?

According to the TTSGenerationResult interface defined in lib/audio/types.ts, the generateTTS() function returns an object with audio: Uint8Array containing raw audio bytes and format: string indicating the codec. Typical formats include 'mp3', 'wav', or 'ogg' depending on the specific provider's output configuration.

Can I use custom TTS endpoints with OpenMAIC?

Yes, the lemonade-tts provider supports OpenAI-compatible API endpoints, allowing integration with custom TTS deployments or proxy services. Configure it by setting providerId: 'lemonade-tts' and providing the custom base URL and API key in the configuration object passed to generateTTS().

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}