HybridAIOne Speech Detect Language
Use the native audio_transcribe tool with action: "detect-language" when
the user asks what language is spoken in an audio clip or when a later
transcription workflow needs a routing decision.
Pass audio as a current attachment filename/ref, /workspace path,
/discord-media-cache path, /uploaded-media-cache path, or HTTPS media URL.
Use provider: "auto" unless the user requests a specific provider.
Return the detected language, provider, duration, cost, and any warnings. Do
not present the full transcript unless the user asked for transcription.