use-local-whisper

To transcribe WhatsApp voice notes locally, this Skill configures whisper.cpp to process audio messages offline, replacing cloud APIs with an on-device pipeline. It requires downloading audio through Baileys-based channel logic and processing it with a local GGML model file.

Running offline voice transcription with whisper.cpp requires the voice-transcription skill, the whisper.cpp binary, ffmpeg, a GGML model file, and launchd PATH configuration for the local binary. The setup verifies these dependencies before applying the local processing change.

Yes, local whisper.cpp transcription works entirely offline without network dependency. It processes WhatsApp audio messages on-device using Apple Silicon, ensuring private voice-note workflows without external data exposure or API costs.

To switch WhatsApp audio transcription from cloud to local, the Skill merges the local-whisper branch, rebuilds the project, updates the service configuration, verifies the model files and launchd PATH, and provides troubleshooting guidance for the restart steps.

Local whisper.cpp setup may fail if the binary, ffmpeg, model files, or launchd PATH are incorrectly configured. The Skill provides deterministic setup verification and troubleshooting guidance to ensure the service PATH and build steps are correctly applied for on-device processing.

Yes, you can use whisper.cpp for private voice transcription on Apple Silicon. This Skill applies on-device processing to WhatsApp voice messages, ensuring audio data remains completely offline and private without relying on external cloud transcription services.