openclaw - 💡(How to fix) Fix feat(tts): support MiMo v2.5 TTS VoiceDesign [1 participants]

Feature request

Please add support for Xiaomi MiMo v2.5 TTS VoiceDesign in OpenClaw's TTS provider configuration.

Background

OpenClaw can already use the Xiaomi TTS provider with mimo-v2.5-tts and a fixed voice value, for example:

{
  "messages": {
    "tts": {
      "auto": "tagged",
      "provider": "xiaomi",
      "providers": {
        "xiaomi": {
          "apiKey": "${XIAOMI_API_KEY}",
          "model": "mimo-v2.5-tts",
          "voice": "Mia",
          "format": "mp3"
        }
      }
    }
  }
}

However, the built-in/preset voices can sound unnatural for some use cases. MiMo v2.5 TTS appears to support a VoiceDesign style/custom voice design capability, which would allow users to describe or configure a desired voice instead of only selecting one of the preset voice IDs.

Requested behavior

Allow OpenClaw's Xiaomi TTS provider to pass through MiMo v2.5 TTS VoiceDesign parameters from config, e.g. one of these shapes or a similar schema:

{
  "messages": {
    "tts": {
      "provider": "xiaomi",
      "providers": {
        "xiaomi": {
          "apiKey": "${XIAOMI_API_KEY}",
          "model": "mimo-v2.5-tts",
          "voiceDesign": {
            "prompt": "A warm, natural young female Chinese voice, soft but clear, conversational, not overly dramatic"
          },
          "format": "mp3"
        }
      }
    }
  }
}

or, if the upstream API expects a different request field name, expose a documented config field and map it internally.

Why this is useful

• Preset voices like Mia may sound odd or unsuitable in real chat usage.
• VoiceDesign would let users create a voice closer to their assistant persona or product tone.
• It avoids requiring users to wait for new preset voices or hardcode custom API payloads outside OpenClaw.

Suggested implementation notes

• Keep existing voice behavior unchanged for backward compatibility.
• If voiceDesign is present, pass the VoiceDesign payload to the Xiaomi MiMo v2.5 TTS API request.
• Validate/guard this feature to models that support it, such as mimo-v2.5-tts.
• Document the config schema in the TTS docs/config reference.

Thanks!