Whisper — локальная транскрипция аудио

★ 7.5 · general

openai-whisper is a Claude Code skill that runs local audio transcription via the Whisper CLI with no API key required. It supports a range of models from tiny to large — defaulting to turbo on install — with model files downloaded to ~/.cache/whisper on first use. Core operations include transcribing audio files to txt or srt format and translating speech directly to English using the --task translate flag, with --model letting you trade speed for accuracy. A practical fit for offline transcription of meetings, interviews, and podcasts, or for embedding speech-to-text into local automation scripts.