How do I enable WhatsApp voice message transcription in openclaw.json?

[visual yarrow]

'm on OpenClaw 4.15, using the WhatsApp channel. When I send a voice message, my agent sees media:audio as a placeholder instead of transcribing it. My OpenAI API key is already configured (openai:default profile). What's the correct config to enable audio transcription?

[scarlet sable]

To help others find answers, you can mark your question as solved via Right click solution message -> Apps -> ✅ Mark Solution

[prisma urchin]

Use the media-understanding audio config, not the WhatsApp channel config.

Minimal openclaw.json:

{
  tools: {
    media: {
      audio: {
        enabled: true,
        models: [
          {
            provider: "openai",
            model: "gpt-4o-mini-transcribe",
            profile: "default"
          }

[prisma urchin]

        ]
      }
    }
  }
}

If you want the higher-accuracy OpenAI transcription model, use:

model: "gpt-4o-transcribe"

Then restart the gateway and test with another WhatsApp voice note. On success the agent should receive an [Audio] block with Transcript:, and {{Transcript}} is available to templates. If you want OpenClaw to also echo the transcript back into WhatsApp, add echoTranscript: true.

Docs: