tools.media.audio does not work
Hier, direkt copy-pastebar:
───
Hey! I'm running OpenClaw 2026.2.26 on Linux (LXC container) with Telegram channel integration. I've configured tools.media.audio with Google Gemini Flash to transcribe incoming voice messages:
"tools": {
"media": {
"audio": {
"enabled": true,
"scope": {
"default": "deny",
"rules": [{ "action": "allow", "match": { "chatType": "direct" } }]
},
"models": [{ "provider": "google", "model": "gemini-3-flash-preview" }]
}
}
}
Every time I send a Telegram voice message (OGG/Opus) to my bot, the Telegram polling loop for that bot account silently dies — no log entry, no error, just stops receiving all messages (including text). Gateway restart restores it until the next voice message.
I'm aware of #7899, #17101, #22554 which describe transcription not working — but my case is more severe: the polling loop crashes. I suspect because my GEMINI_API_KEY is present in the gateway environment, the audio processing actually starts (unlike users without a key), fails with an unhandled exception, and kills the loop.
Questions:
1. Is this a known issue? Is there a fix in progress or a safe workaround to receive voice messages without crashing the polling loop?
2. Is tools.media.audio for Telegram fundamentally broken right now, or am I missing something in the config?
3. Any ETA on a fix for applyMediaUnderstanding not being called for Telegram voice?
Currently have audio.enabled: false as a workaround.
───
Hey! I'm running OpenClaw 2026.2.26 on Linux (LXC container) with Telegram channel integration. I've configured tools.media.audio with Google Gemini Flash to transcribe incoming voice messages:
"tools": {
"media": {
"audio": {
"enabled": true,
"scope": {
"default": "deny",
"rules": [{ "action": "allow", "match": { "chatType": "direct" } }]
},
"models": [{ "provider": "google", "model": "gemini-3-flash-preview" }]
}
}
}
Every time I send a Telegram voice message (OGG/Opus) to my bot, the Telegram polling loop for that bot account silently dies — no log entry, no error, just stops receiving all messages (including text). Gateway restart restores it until the next voice message.
I'm aware of #7899, #17101, #22554 which describe transcription not working — but my case is more severe: the polling loop crashes. I suspect because my GEMINI_API_KEY is present in the gateway environment, the audio processing actually starts (unlike users without a key), fails with an unhandled exception, and kills the loop.
Questions:
1. Is this a known issue? Is there a fix in progress or a safe workaround to receive voice messages without crashing the polling loop?
2. Is tools.media.audio for Telegram fundamentally broken right now, or am I missing something in the config?
3. Any ETA on a fix for applyMediaUnderstanding not being called for Telegram voice?
Currently have audio.enabled: false as a workaround.
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
Similar Threads
Was this page helpful?
