@itou_tool/minimax-speech-provider-plugin

Standalone MiniMax speech provider plugin for OpenClaw.

This plugin registers a TTS provider with:

• plugin id: minimax-speech-provider-plugin
• speech provider id: minimax
• npm package: @itou_tool/minimax-speech-provider-plugin

It is designed as an independent plugin package that can live outside the main OpenClaw repo and later be published to npm.

Package metadata for npm publishing lives in package.json, release notes in CHANGELOG.md, and a short publish checklist in PUBLISHING.md.

For compatibility with OpenClaw v2026.3.23-2, the plugin keeps its runtime entry self-contained and does not rely on host-side openclaw/plugin-sdk/* module aliasing while loading.

No build step is required for release. The published package intentionally ships the source .ts files that OpenClaw loads at runtime.

Install

Local development:

openclaw plugins install -l ./plugins/minimax-speech-provider-plugin

From npm later:

openclaw plugins install @itou_tool/minimax-speech-provider-plugin

Restart the gateway after install or config changes.

Config

Set the plugin entry under plugins.entries.minimax-speech-provider-plugin and select minimax as the active TTS provider:

{
  plugins: {
    enabled: true,
    load: {
      paths: ["./plugins/minimax-speech-provider-plugin"],
    },
    entries: {
      "minimax-speech-provider-plugin": {
        enabled: true,
        apiKey: "your-minimax-api-key",
        config: {
          baseUrl: "https://api.minimax.io",
          model: "speech-2.8-hd",
          voiceId: "English_expressive_narrator",
          audioFormat: "mp3",
          languageBoost: "auto",
          speed: 1,
          volume: 1,
          pitch: 0,
          timeoutMs: 60000,
        },
      },
    },
  },
  messages: {
    tts: {
      provider: "minimax",
    },
  },
}

If you use the China endpoint, set:

{
  plugins: {
    entries: {
      "minimax-speech-provider-plugin": {
        config: {
          baseUrl: "https://api.minimaxi.com",
        },
      },
    },
  },
}

apiKey also falls back to MINIMAX_API_KEY.

What v1 supports

• static plugin config
• normal speech synthesis via MiniMax t2a_v2
• optional voice listing via MiniMax get_voice
• mp3, wav, flac, ogg, and opus output

Current limitation

OpenClaw core does not yet parse MiniMax-specific [[tts:...]] override fields. So this plugin works with static config, but not with MiniMax-only per-message directive overrides such as MiniMax-specific voice_id or emotion controls.