Bug type

Behavior bug (incorrect output/state without crash)

Beta release blocker

No

Summary

Summary

openclaw infer audio transcribe returns:

• Error: No transcript returned for audio: <file>

for multiple providers on valid local audio files.

This reproduces with:

• Deepgram
• Groq
• OpenAI

It also reproduces on:

• original Telegram voice note (.ogg / opus)
• the same file converted locally to .wav via ffmpeg

This makes it look like a shared OpenClaw audio transcription pipeline issue rather than a single provider problem.

Environment

• OpenClaw: 2026.4.10
• Host: Ubuntu VM in VirtualBox
• Channel in use: Telegram
• Audio file tested locally from:
  • ~/.openclaw/media/inbound/...ogg
• Converted test file:
  • /tmp/test.wav

What I configured

I tested provider auth through environment variables.

openclaw models status shows provider env auth is visible:

• deepgram → effective=env
• groq → effective=env
• openai-codex → OAuth profile

However, openclaw infer audio providers shows:

• deepgram: "available": true, "configured": false
• groq: "available": true, "configured": false
• openai: "available": true, "configured": true

That mismatch may be relevant.

Reproduction

1. Test on local Telegram voice note

openclaw infer audio transcribe \
  --file /home/vboxuser/.openclaw/media/inbound/file_5---a4beb8cf-e807-471e-a885-d71eed4d7032.ogg \
  --model deepgram/nova-3 \
  --json


### Steps to reproduce

NOT_ENOUGH_INFO

### Expected behavior

Transcribe success

### Actual behavior

NOT_ENOUGH_INFO

### OpenClaw version

2026.4.10

### Operating system

Ubuntu 24.04

### Install method

npm global

### Model

Codex 5.4

### Provider / routing chain

infer audio transcribe -> provider selection -> STT provider call -> empty/no transcript result

### Additional provider/model setup details

_No response_

### Logs, screenshots, and evidence

```shell

Impact and severity

No response

Additional information

No response