Connect Kitta Audio TTS to OpenClaw

OpenClaw's OpenAI TTS provider can override its Base URL and treats a non-default URL as an OpenAI-compatible endpoint. Use it for explicit /tts commands or automatic voice replies.

Add the provider settings

Add this to your OpenClaw configuration:

{
  messages: {
    tts: {
      auto: 'always',
      provider: 'openai',
      providers: {
        openai: {
          baseUrl: 'https://fishaudio.org/v1',
          apiKey: 'YOUR_API_KEY',
          model: 'fishaudio-s21pro-flash',
          speakerVoice: '00a1b221-6137-4b73-ad62-b0cbce134167',
          responseFormat: 'mp3',
        },
      },
    },
  },
}

You may instead set OPENAI_TTS_BASE_URL=https://fishaudio.org/v1 to override only the TTS endpoint. Prefer environment variables or OpenClaw's secret management over storing a key in source-controlled configuration.

Test the connection

Restart or reload OpenClaw, then run:

/tts status
/tts audio Hello, this is a Kitta Audio connection test.

After the audio plays, replace speakerVoice with your own Voice ID. Set auto to off if you only want speech when /tts audio is used.

Troubleshooting

• Requests still reach OpenAI: ensure baseUrl is nested under messages.tts.providers.openai.
• Invalid voice: speakerVoice must contain a Kitta Audio Voice ID.
• No native voice bubble: channels handle MP3 and Opus differently; first confirm that the audio attachment plays.
• 401/402: verify the API key and separate API credit balance.

Related documentation

• Find a Voice ID
• OpenAI-Compatible TTS
• Official OpenClaw OpenAI provider guide
• Official OpenClaw TTS guide