Telegram: voice/audio-only messages don't trigger a response

Symptoms

• Sending a Telegram voice note (audio-only) produces no reply.
• If you add any text (for example, ? + voice note), the bot replies and the transcription works.

Cause

Audio-only messages need a transcription step to produce a usable text prompt. If inbound audio transcription is disabled (or blocked by limits/scope), the gateway may treat the message as “no text”, so it never triggers a normal reply flow.

Fix

1) Enable inbound audio transcription (recommended)

In your config, enable tools.media.audio and pick a provider (or let auto-detection work).

Example (OpenAI transcription):

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

Restart the gateway after changes.

2) Ensure media limits allow voice notes

• Telegram media cap: channels.telegram.mediaMaxMb (default 5MB)
• Audio transcription cap: tools.media.audio.maxBytes (default 20MB)

If your voice notes are larger than the Telegram cap, they may never be downloaded/transcribed.

3) Probe and check logs

On the gateway host:

openclaw channels status --probe
openclaw logs --follow

Look for audio transcription messages (or errors about media size/timeout/auth).

Verify

• Send an audio-only voice note; the bot replies with the transcript-based answer.
• The logs show an audio transcription decision/output.