Audio auto-transcription not triggering on Telegram voice notes
✅Solved
Running OpenClaw 2026.2.9 on macOS (arm64), Telegram channel. Voice notes arrive with <media:audio> tag but the transcription pipeline never runs — no audio-related entries in logs at all.
Config:
{
"tools": {
"media": {
"audio": {
"enabled": true,
"models": [
{ "provider": "openai", "model": "gpt-4o-mini-transcribe" }
]
}
}
}
}
Environment:
• OPENAI_API_KEY is set and working (manual Whisper API calls succeed)
• whisper Python CLI is on PATH
• Gateway restarted after config change
• Telegram voice notes are .ogg (opus codec), files download to ~/.openclaw/media/inbound/
What works: Manually running the openai-whisper-api skill script transcribes the files perfectly. The config just doesn't trigger the automatic pipeline.
What I've tried:
• config.patch + SIGUSR1 restart
• Full openclaw gateway restart
• Checked logs — zero audio/transcription/media entries around message receipt
Any ideas? Am I missing a config key, or is there a known issue with the Telegram provider + audio pipeline?
Config:
{
"tools": {
"media": {
"audio": {
"enabled": true,
"models": [
{ "provider": "openai", "model": "gpt-4o-mini-transcribe" }
]
}
}
}
}
Environment:
• OPENAI_API_KEY is set and working (manual Whisper API calls succeed)
• whisper Python CLI is on PATH
• Gateway restarted after config change
• Telegram voice notes are .ogg (opus codec), files download to ~/.openclaw/media/inbound/
What works: Manually running the openai-whisper-api skill script transcribes the files perfectly. The config just doesn't trigger the automatic pipeline.
What I've tried:
• config.patch + SIGUSR1 restart
• Full openclaw gateway restart
• Checked logs — zero audio/transcription/media entries around message receipt
Any ideas? Am I missing a config key, or is there a known issue with the Telegram provider + audio pipeline?
Solution
Your
A few concrete checks that tend to surface the real issue fast:
1) Confirm the running Gateway is actually seeing your config
(If
2) Confirm the Gateway service can see OpenAI credentials
If OpenAI isn’t “available” there, the fix is to put the key somewhere the service can load it:
- put
tools.media.audiowhisperA few concrete checks that tend to surface the real issue fast:
1) Confirm the running Gateway is actually seeing your config
(If
gateway status2) Confirm the Gateway service can see OpenAI credentials
If OpenAI isn’t “available” there, the fix is to put the key somewhere the service can load it:
- put
OPENAI_API_KEY=...~/.openclaw/.env
The AI that actually does things. Emails, calendar, home automation - all from your favorite chat app. New shell, ready to help. The lobster way.
134,426Members
Resources
Was this page helpful?
