openai-whisper-api

Once installed, the skill exposes a shell script that wraps OpenAI's audio transcription API using curl. Here's the basic flow:Point the script at your audio file: transcribe.sh /path/to/audio.m4aThe script sends the file to OpenAI's /v1/audio/transcriptions endpoint using your API key.The transcript is saved alongside the input file as .txt by default, or to a custom path using --out.You can tune the request with flags like --language en to improve accuracy, --prompt to provide context (e.g. speaker names), or --json to receive the raw API JSON response.

Simple CLI Interface — Transcribe audio with a single command; no boilerplate code needed.Language Hints — Pass --language to guide Whisper toward a specific spoken language for better accuracy.Context Prompting — Use --prompt to inject speaker names or domain vocabulary, improving transcript quality.Flexible Output — Save transcripts as plain text (default) or raw JSON with the --json flag.Custom Output Path — Specify any destination file with --out /path/to/output.txt.Configurable Model — Select the Whisper model via --model (defaults to whisper-1).

OpenAI API Key (required) — Authenticates requests to OpenAI's audio transcription service. Set the OPENAI_API_KEY environment variable, or configure it in ~/.openclaw/openclaw.json. Supports: OpenAI.

Meeting & Interview Transcription — Convert recorded meetings or interviews into searchable text transcripts automatically.Content Repurposing — Turn podcast episodes or voice memos into written content ready for editing or publishing.Automated Pipelines — Integrate the script into CI/CD or data pipelines to batch-transcribe audio files without manual effort.Accessibility — Generate text versions of audio content to improve accessibility for hearing-impaired audiences.