whisnap

What This Skill Does

The whisnap skill provides a robust macOS command-line interface for the Whisnap transcription engine. It allows users to leverage either high-performance local Whisper models for privacy-focused, offline-capable transcription or the powerful Whisnap Cloud service for demanding workloads. By integrating directly with the Whisnap macOS application, the skill ensures that your models, authentication, and settings are synchronized between the desktop GUI and your terminal environment. Whether you are dealing with WAV, MP3, FLAC, M4A, OGG, or common video formats like MP4 and MKV, whisnap streamlines the conversion of audio and video into structured, timestamped text data.

Installation

To install this skill, run the following command in your terminal: clawhub install openclaw/skills/skills/neolio42/whisnap. Before usage, ensure the Whisnap desktop application is installed. Navigate to the application Settings, select Advanced, and enable the CLI feature to automatically create the symlink at /usr/local/bin/whisnap. Additionally, verify that you have downloaded at least one Whisper model within the application dashboard to enable local processing.

Use Cases

• Transcription Automation: Incorporate transcription into your development workflows by piping output from meetings or recorded audio directly into text-based analysis tools.
• Content Creation: Quickly transcribe video files (MP4, MOV) to generate captions, transcripts, or searchable archives for your creative projects.
• Privacy-First Workflows: Utilize local model processing for sensitive meetings or recordings where data must remain strictly on the local machine.
• Cloud-Powered Scaling: Offload long audio files to the Whisnap Cloud for faster processing without taxing your local CPU or GPU resources.

Example Prompts

1. "Use whisnap to transcribe my latest meeting recording meeting_final.mp4 and output the results as a JSON file so I can parse the timestamps."
2. "Can you help me transcribe interview.wav using the small-q5_1 model to ensure the output is concise?"
3. "Run a transcript of the audio file in my downloads folder using the cloud processing flag and show me any diagnostic errors if they occur."

Tips & Limitations

• JSON Integration: Use the --json flag for all programmatic tasks. It provides rich metadata, including segment start and end times, which is essential for building custom players or search indexes.
• Troubleshooting: Always check stderr for diagnostics by adding the -v flag. If a file fails, ensure the path is absolute or relative to the current working directory.
• Authentication: Cloud mode requires an active sign-in within the Whisnap app. If transcription fails with a 401-like error, verify your account status in the desktop client.
• Scope: The CLI exclusively supports Whisper-based architectures; Parakeet models are not compatible via this command-line interface.